Reciprocity, Negotiation, and the Choice of Structurally

Home , Generalized exchange

Social Psychology Quarterly 2016, Vol. 79(2) 115–135 Should I Stay or Should Ó American Sociological Association 2016 DOI: 10.1177/0190272516641392 I Go? Reciprocity, Negotiation, http://spq.sagepub.com and the Choice of Structurally Disadvantaged Actors to Remain in Networks

Scott V. Savage1 and Zachary L. Sommer2

Abstract Drawing on existing theories of social exchange as well as self-categorization theory, we consider how two forms of direct exchange influence whether structurally disadvantaged actors choose to stay in the micro-structures that disadvantage them. We posit that (1) the exit opportunity is more likely to result in disadvantaged actors coming to view their network as a group if there has been a history of reciprocal, as opposed to negotiated, exchange and (2) this psychological group formation should account for disadvantaged actors disproportionately choosing to remain in reciprocal exchange networks. We also consider whether the information actors have about the alternative network affects this choice. Findings from two laboratory experiments generally support our argument that for disadvantaged actors, psychological group formation mediates the relationship between exchange form and staying in networks.

Keywords social exchange, self-categorization theory, network dynamics, group formation

The stability of network exchange struc- network expansion. Cook and Gillmore tures that privilege some and disadvan- (1984) demonstrated that given the tage others is the subject of much theoriz- chance, disadvantaged actors will use ing and research (e.g., Cook and Gillmore coalition formation to improve their bar- 1984; Dogan et al. 2009; Leik 1992; Rand, gaining power. Similarly, Leik (1992) Arbesman, and Christakis 2011; Willer described how network expansion alters and Willer 2000). This scholarship often power relations by changing dependen- builds on Emerson’s (1972) idea that cies between actors. He also stressed the power imbalanced networks should move toward equality as disadvantaged actors 1 attempt to improve their situation by University of Houston, TX, USA 2University of California, Riverside, CA, USA either altering their or their exchange partner’s values or changing the struc- Corresponding Author: Scott V. Savage, Department of Sociology, tural conditions of the network. University of Houston, 474 Philip G. Hoffman Hall, The two structural changes that Emer- Houston, TX, 77204, USA. son identified are coalition formation and Email: [email protected]

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 116 Social Psychology Quarterly 79(2) need for research on what motivates indi- integrating force (e.g., Molm, Collett, viduals to add or delete linkages in order and Schaefer 2007; Molm, Whitham, and to understand ‘‘which power structures Melamed 2012) that can motivate the are self-perpetuating and enhancing vs. development of psychological contracts random consequences of change processes or beliefs in the mutual obligation of self vs. the source of their destruction’’ (Leik and other (Rousseau 1989; Rousseau and 1992:322). Willer and Willer (2002) Tijoriwala 1998). Building on this insight, extended this idea by using network we investigate how different exchange exchange theory to explain how rational, histories shape whether disadvantaged fully informed actors might improve or actors choose to stay in the very networks defend their power by adding or deleting that disadvantage them. We theorize (1) linkages to others. They concluded that that the exit opportunity is more likely because adding links generally benefits to result in disadvantaged actors coming those in low power positions, rational to view their network as a group if there actors in low power positions should try has been a history of reciprocal, as to add linkages until networks become opposed to negotiated, exchange and (2) power balanced. To counteract this and that this psychological group formation maintain power, those in power advan- should affect the exit choice. Thus, we taged positions should respond by limit- add to the extant research on network ing the ability of others to seek out new stability by considering the underlying connections. psychological processes that influence One way of accomplishing this is by the staying behaviors of those who have creating a ‘‘with us or against us’’ sce- the strongest material incentive to leave. nario, whereby individuals must choose between existing and alternative sets of THEORY network ties. Although extreme, such sit- Social Exchange uations do occur. Employers often demand the full commitment of employ- The social exchange perspective views ees, forcing them to choose between the human interaction as a series of resource current employer and alternative employ- exchanges between actors (Homans ers (Hirschman 1970; Kmec 2007). Chil- 1974). A resource in social exchange is dren sometimes do something similar any capability or object that is held by when constructing friendship groups. one actor and valued by another. Adler and Adler (1995), for example, Exchange relationships form when the showed that joining a more popular resources to be had from interdepen- friendship network sometimes means cut- dently exchanging with others outweigh ting ties with one’s current set of friends. independent action (Thibaut and Kelley That it is possible to sever ties with one 1959). That is, in the pursuit of valued network in order to join a new one, how- resources, actors enter into recurring ever, doesn’t address why those disadvan- exchange relations with those who can taged by the structure of the network provide said benefits. might choose to remain in it. Resource exchanges can take various One possible explanation for this forms. When the outcomes of exchange hinges on gift giving and its underlying depend on the behaviors of actors who norm of reciprocity. Ongoing research by exchange with one another, an exchange social exchange theorists finds that apart relation is said to be direct, and from the benefits to be had, the structure there are two forms of direct exchange of reciprocal exchange can serve as an relations: negotiated and reciprocal.

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 Should I Stay or Should I Go? 117

Negotiated exchanges are those where an individual wants to buy a car and he actors bargain over the terms of the chooses to buy it from a certain dealer, exchange and engage in an exchange this necessarily precludes buying the car transaction only after the terms of the from all other dealers. Power imbalances exchange have been agreed on, such as in such networks arise due to inequalities those associated with buying a house or in dependence. When network alterna- a car. Typically, social exchange research- tives result in one actor being less depen- ers rule out opportunism by assuring that dent on another actor for resources than actors know about the quality of the the other actor is on him or her, a power resources to be exchanged (for an excep- imbalanced exchange relation exists, tion, see Kollock 1994) and making agree- with the less dependent actor experienc- ments binding (e.g., Cook et al. 1983). ing a power advantage (Emerson 1972). Negotiated exchanges are characteristic We consider why those disadvantaged of traditional markets, as exemplified by by these structural conditions sometimes sports teams bargaining with one another opt to stay in them and whether this over the terms of a trade and players bar- choice is influenced by the form of gaining with sports teams over the terms exchange. Key to our argument is the of a contract (for an example, see Lawler, mechanism of psychological group forma- Yoon, and Thye 2009). tion or the process by which someone Reciprocal exchanges are those where comes to view himself or herself as part actors independently initiate an exchange of a group of three or more actors charac- by providing a benefit to another, for terized by increased cooperation and example, swapping birthday presents. A reduced levels of self-interest (Hogg lack of negotiation and binding agree- 1992; Lawler and Yoon 1996). In the fol- ment means that benefits flow unilater- lowing, we explain why a history of recip- ally and that there is no assurance that rocal exchange should make disadvan- giving will result in another’s reciproca- taged actors more likely to undergo tion, which makes these exchanges risk- psychological group formation when ier than negotiated exchanges (e.g., given the exit opportunity and thus stay Cheshire, Gerbasi, and Cook 2010; Molm in a negatively connected, power imbal- 1994; Molm et al. 2007). Reciprocal anced network. exchange is an important characteristic of the network form of organization (Powell 1990) and a prominent feature of Form of Exchange, Group Formation, the 23 women’s better-dress firms that and Disadvantaged Actors Staying Uzzi (1997) studied. in Networks Although both forms of exchange occur The structure of exchange can have directly between two actors, these actors important consequences for the bonds may be embedded in more complex net- that develop between actors (e.g., Kuwa- works of exchange relations. Such bara 2011; Lawler, Thye, and Yoon exchange structures consist of the connec- 2008; Molm et al. 2007; Willer, Flynn, tions between a set of exchange actors and Zak 2012). Thye, Lawler, and Yoon and shape the mutual dependence of (2011), for example, have shown that by actors. Negatively connected network influencing exchange frequency, the posi- exchange structures are those structures tive emotions that result from exchange, where exchange in one relation precludes and perceptions of network cohesion, two exchange in another on any given dimensions of exchange networks—the exchange opportunity. For example, if network-wide potential for exchange and

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 118 Social Psychology Quarterly 79(2) the inequality of structural power—can networks expressing greater personal affect whether actors who negotiate over attractions for particular exchange part- the distribution of a resource are likely ners than those embedded in negotiated to come to view their exchange network exchange networks (Molm et al. 2007). as a group. Willer et al. (2012) offer a dif- This finding reinforces Hogg and Turner’s ferent insight, showing that generalized (1985) claim that while personal attrac- exchange systems, which involve indirect tions can serve as the basis for group for- giving and receiving of benefits such that mation, they are unlikely to influence no individual receives benefits directly group formation if the relationships are from the person to whom they gave, foster idiosyncratic and personal. So even greater group identification than direct though reciprocal exchange results in exchange systems and that this effect is stronger bonds between pairs of exchange mediated by the number of benefits partners, namely, personal attractions, exchanged. than negotiated exchange (Molm, Collett, This finding departs from the predic- and Schaefer 2006; Molm et al. 2007, tions of the affect theory of social 2012; Molm, Takahashi, and Peterson exchange (Lawler 2001; Lawler et al. 2000), actors in reciprocal exchange net- 2008), which posits that because the joint- works appear no more likely to view their ness of exchange is not as explicit in gen- networks as groups. eralized exchange as it is in the two direct We argue that the exit opportunity forms of exchange and in productive changes this by introducing into the situ- exchange, actors in generalized exchange ation a categorical distinction that can systems should be less inclined to develop serve as the basis for an ingroup/outgroup a sense of shared responsibility for distinction. Self-categorization theory exchange outcomes, attribute emotions assumes a social situation will motivate to the social unit, and thus form a sense psychological group formation if (1) an of cohesion with the network.1 While individual’s past experiences and present empirical research testing this idea shows desires prepare him or her to use a per- that generalized exchange produces less ceived categorical distinction and (2) the network cohesion and group attachments individual thinks categorization mini- than the other forms of exchange, it does mizes intraclass differences relative to not show a significant difference in group interclass differences in expected or ste- attachments for the two direct forms of reotypical ways (Turner et al. 1994; exchange that we study (Lawler et al. Turner and Reynolds 2012). 2008). A history of reciprocal exchange is Similarly, Molm and colleagues find more likely to satisfy these conditions that actors in negotiated and reciprocal for disadvantaged actors than a history exchange networks rarely describe their of negotiated exchange. Because disad- set of exchange relations as a ‘‘group, vantaged actors experience negotiation a team, working together’’ (Molm et al. as more conflict laden and less fair than 2007:234). Rather, they view the bonds reciprocity (Molm et al. 2003, 2006), a his- that exist between exchange partners as tory of negotiation should result in disad- distinct and isolated units, with those vantaged actors being less prepared to embedded in reciprocal exchange use the categorical distinction as the basis for psychological group formation. More- 1Productive exchanges occur when people uni- over, because negotiation clashes with laterally contribute to the production of a joint the friendship norms of welfare and fair- good; coauthoring a paper is an example. ness (Kurtzberg and Medvec 1999), it

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 Should I Stay or Should I Go? 119 should be easier for disadvantaged actors Consequently, we contend that following to minimize intraclass differences and the exit opportunity, disadvantaged view exchange partners as behaving in actors will rely on their personal attrac- ‘‘groupy’’ ways if they had previously tions as a basis for group formation. engaged in reciprocal exchange. Thus, Because personal attractions differ across we contend that when presented with the two forms of direct exchange, personal the opportunity to defect to another net- attractions should mediate the relation- work, those in reciprocal exchange net- ship between exchange form and psycho- works will be more likely to use the cate- logical group formation. gorical distinction made evident by the presence of an alternative network to Hypothesis 2: When disadvantaged undergo psychological group formation. actors become aware of the opportunity to join an alternative network, Hypothesis 1: The exit opportunity will their personal attractions for their be more likely to result in psychologi- respective exchange partners will pos- cal group formation for disadvantaged itively affect whether they view their actors in reciprocal exchange net- existing network as a group. works than in negotiated exchange Hypothesis 3: Personal attraction will networks. mediate the relationship between exchange form and psychological group formation. We also suspect that perceptions of personal attraction will mediate this rela- In addition to personal attractions, the tionship. How resources are exchanged amount of information disadvantaged across the two forms of direct exchange actors have about the alternative may results in reciprocal exchange being risk- affect group formation. Hogg (2012) con- ier, less conflict laden, and more symboli- tends that the desire to reduce subjective cally valuable than negotiated exchange, uncertainty about the social world moti- and this results in greater feelings of per- vates individuals to think of themselves sonal attraction for specific exchange as group members so that they can form partners in reciprocal exchange networks expectations about what they might gain (Molm et al. 2007), with these effects from their interactions. Thus, the amount holding for those disadvantaged by the of information an actor has about an exchange structure (Molm et al. 2000, exchange network should affect percep- 2012). These differences in personal tions of uncertainty (Savage and Berg- attraction, which we define as perceptions strand 2013) and consequently psycholog- of social unity with a specific exchange ical group formation. A complete lack of partner, become relevant when actors information about the alternative makes become aware of the opportunity to join it impossible for disadvantaged actors to an alternative network insofar as they compare the present network with the provide ‘‘a cognitive criterion for common alternative, maximizing the uncertainty category membership’’ (Hogg and Turner in their subjective worlds. Interestingly, 1985:61). This is likely because positive the provisioning of additional, albeit personal attractions amplify perceptions incomplete, information about the alterna- of similarity (e.g., Backman and Secord tive network does not improve the ability 1962; Hogg and Turner 1985), thereby to determine whether the new exchange resulting in the minimization of intra- network will provide greater or lesser ben- class differences relative to interclass dif- efits. Still, the extra information might ferences (Turner and Reynolds 2012). result in actors believing they are better

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 120 Social Psychology Quarterly 79(2) able to ascertain the benefits they will get of random assignment in concert with var- from the alternative, thereby lessening ious experimental controls ruled out the perceptions of uncertainty and the motiva- possibility of systematic variation across tion for psychological group formation. experimental conditions (Campbell and Thus, we introduce our fourth hypothesis. Stanley 1963) and allowed for a rigorous test of our predictions. Hypothesis 4: Psychological group formation will be greater when the amount of uncertainty about the benefits pro- Design and Participants vided by exchange relations in an Social exchange theory assumes actors alternative exchange network is high value the resources of exchange (Molm rather than low. and Cook 1995). A precondition for partic- That the exit opportunity should differ- ipation in the experiment then was a desire for the resource of exchange. entially affect psychological group forma- Thus, we recruited undergraduate stu- tion across the two forms of direct dents from a large, public university exchange matters because how strongly based on their desire for the resource in an individual identifies with a group the experiment: money. The experiment should affect whether the actor stays in manipulated the form of exchange (recip- it. Previous research demonstrates that rocal vs. negotiated exchange) and the personal attractions are often poor predic- information subjects had about the alter- tors of behavior at units of analysis native network (low vs. high information), beyond the dyad and that a group identity thereby allowing us to evaluate our theo- is a better predictor of behaviors at the retical argument.2 Sixty-four undergrad- network level (Hogg 1992). Thus, whether uate students were randomly assigned to an actor chooses to remain in an existing each of the experimental conditions cre- exchange network or not should depend ated by crossing the form of exchange on whether the actor views the network and the amount of information, so that as a group to which he or she belongs. there were 16 students per condition.3 For control purposes, equal numbers of Hypothesis 5: Psychological group forma- men and women were assigned to each tion group will positively affect 4 whether disadvantaged actors choose condition. to remain in an existing exchange network. 2In addition to the form of exchange and infor- Hypothesis 6: Psychological group forma- mation, Experiment 1 manipulated power by tion will mediate the relationship placing participants in either high or low power between the form of exchange and conditions. While the manuscript focuses on the low power conditions, online Appendix A (avail- staying in an existing network. able at spq.sagepub.com/supplemental) reports the results of a mediation analysis using the We test these hypotheses with data from KHB method to examine the effects of power. two controlled laboratory experiments. This analysis reveals that power moderates the effects of psychological group formation on stay- EXPERIMENT 1 ing behavior; the effects are strongest for those in low power conditions. This finding strengthens Experiment 1 involved undergraduate stu- our confidence in our argument. dents who earned money by exchanging 3Sixty-seven undergraduate students actually with two computer-simulated actors. The participated in these conditions. We excluded data from three subjects because they did not amount earned depended on participants’ believe they were interacting with real people. exchanges with partners. Implementation 4Sensitivity analyses found no gender effects.

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 Should I Stay or Should I Go? 121

also learned they would be interacting with their partners over a series of exchange opportunities and that for any one opportunity they would only be able to exchange with one partner.5 Subjects were aware of the potential benefits they could receive from each partner on each Figure 1. Diagram of the Network That exchange opportunity and thus were Appeared on Subjects’ Screens Note: Subjects were told that they were W and that aware that they had one high value (X) they could exchange with X and Z. and one low value (Z) partner. The greater benefits potentially provided by X combined with the fact that X also had Procedures a high value relation with Y meant subjects were more dependent on X for The experiment involved participants resources than X was on them and as engaging in a series of exchange opportu- a result, at a power disadvantage. They nities with two exchange partners in were unaware of this. what they believed was a fixed, four- After the instructions, subjects partici- person exchange network. While multiple pated in a series of exchange opportuni- subjects could participate in the experi- ties with their two simulated exchange ment at the same time, no two subjects partners. The completion of each encountered one another during the exchange opportunity resulted in partici- experiment. Upon arriving, participants pants receiving information about who were directed to isolated rooms where gave them points and how many points they were told that instructions for the they received. Total earnings were experiment would appear on the com- updated and reported. In order to reduce puter terminal once all participants had equity effects (Cook and Emerson 1978) arrived. and enhance the uncertainty associated Instructions informed subjects that with exchange (Molm et al. 2000), sub- they would be interacting with other stu- jects did not receive information about dents and the amount of money they the profits earned by their exchange earned would depend on these interac- partners. tions. Unbeknownst to participants, We divided the exchange opportunities exchange partners were simulated actors into three phases, with the number of designed to mimic real exchange behav- exchange opportunities in each phase ior. Simulated actors were used as a varying by the phase and the form of control to ensure that subjects had comparable earnings across the forms of 5Subjects exchanged in negatively connected exchange (mean = 1,235.88 in reciprocal; networks (Cook and Emerson 1978). To operationalize a negatively connected network under mean = 1,216.13 in negotiated; t test62 = .60, ns). The instructions also taught par- conditions of negotiated exchange, actors could make an agreement with one, and only one, ticipants how to use their computers to exchange partner on any given exchange opportu- make exchanges. Participants were told nity. To do so in reciprocal exchange conditions, that they would have two exchange part- actors could only give to one exchange partner on any particular exchange opportunity. ners who each had a common alternative 6 partner. A diagram of this exchange net- These exchange networks are weak power network structures, because ‘‘no position ...is work appeared on subjects’ screens and assured of being able to exclude another without can be seen in Figure 1. Participants cost’’ (Markovsky et al. 1993:202).

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 122 Social Psychology Quarterly 79(2) exchange. After completing 62 percent of amount of benefits when making requests, exchange opportunities, the first phase thereby masking how many points their stopped, and subjects were informed that partner received from exchange. On any they were partway through the experiment given exchange opportunity, subjects nego- and asked to answer a few questions about tiated for up to four rounds and were able theirdyadicrelationshipsandpersonal to make offers, make counteroffers, or attractions for each exchange partner accept offers on each round. Each round (Molm et al. 2007). After answering these had subjects request how many points questions, the second exchange phase they wanted to receive from their exchange began. It stopped after a predetermined partners, and the computer converted number of exchange opportunities, with these requests into offers to the partners. the number varying by condition. Here sub- Agreements occurred when an actor’s jectsweretoldthattheyhadbeenran- request matched his or her partner’s offer. domly selected for a unique opportunity If an agreement was made, the actor imme- and that at some point in the next part of diately received that amount. If an actor theexperimenttheywouldbeabletointer- failedtomakeanagreement,theactor act with a new set of people should they did not receive points for that exchange want to stop interacting with the people opportunity. with whom they had been interacting. Reciprocal exchange conditions con- The amount of information subjects sisted of a series of exchange opportuni- received about this alternative network ties in which subjects had to select to varied by condition. Subjects then were whom they wanted to give points without asked to answer questions designed to mea- knowing the choices of others. So, on any sure psychological group formation. The particular exchange opportunity, subjects third exchange phase started upon comple- made their selection without knowing if tion of the questionnaire. After a number of anyone was giving to them. Giving points additional exchanges, with the number did not result in the giver losing points varyingbycondition,subjectsreceivedthe from his or her total. At the end of each opportunity to join the alternative network. exchange opportunity, subjects were told The experiment ended with the subject’s who gave to them and how much they choice to either join an alternate network received. All analyses examine the effect or stay in their current network. All sub- of reciprocal exchanges by making negoti- jects were debriefed and paid. ated exchange the reference category. Steps were taken to ensure these two forms of exchange were as similar as pos- Manipulations sible. First, more exchange opportunities were run in reciprocal exchange condi- Form of exchange. Participants engaged tions (160) than in negotiated exchange in either reciprocal or negotiated exchange. conditions (80). Each exchange opportu- We manipulated these using exchange set- nity in negotiated exchange conditions tings developed by Molm, Peterson, and allowed for up to four rounds of bargain- Takahashi (1999). Negotiated exchange ing, while each opportunity in reciprocal conditions had subjects bargain with sim- exchange consisted of a single act. Dou- ulated exchange partners over the division bling the number of reciprocal exchange of a pool of benefits. Subjects knew the opportunities approximately equates the range of points they could request from time and effort spent by participants their exchange partners but did not across these two forms of exchange. know they were simply dividing a fixed Exchange opportunities were distributed

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 Should I Stay or Should I Go? 123 across the three exchange phases with Subjects had one high power simulated phase 1 consisting of 100 reciprocal exchange partner (the focal partner) and exchange opportunities or 50 negotiated one low power simulated exchange part- exchange opportunities, phase 2 consist- ner (the other partner) who each had ing of 40 reciprocal exchange opportuni- another high power exchange partner in ties or 20 negotiated exchange opportuni- common. For both forms of exchange, ties, and phase 3 consisting of 20 the behaviors of the focal partner were reciprocal exchange opportunities or 10 set to disadvantage the low power partic- negotiated exchange opportunities. ipant, and the behaviors of the other part- Second, the monetary values of points ner were programmed to create, on aver- were adjusted to ensure comparable age, equal value exchanges between the behaviors received comparable earnings low power participant and the simulated across the two forms of exchange. The total actor. Interactions between participants number of points an actor could give to an and the focal partner disadvantaged the exchange partner on any exchange oppor- participant so that he or she received, on tunity in reciprocal exchange was equal average, 5 points for every 7 points the to one-half the points that could be divided simulated actor received from the subject. between actors in the negotiated exchange In the negotiated exchange conditions, setting on each opportunity. this disparity between the participant Amount of information about the exit and the focal partner was created by opportunity. We randomly placed subjects manipulating the number of points in one of two information conditions. In requested by the focal actor. The focal the first condition, subjects were told actor made initial requests of 7, 8, or 9 they had an opportunity to exchange in points, meaning subjects were only being an alternative network but received no offered 5, 4, or 3 points. If this request information about the structural dimen- was not accepted, the request was low- sions of the network. Importantly, the ered by 1 on the next round of negotia- prompt emphasized that subjects could tions and then repeated on all subsequent not predict beforehand whether leaving rounds. For each exchange opportunity, would result in better or worse earnings. the high power simulated actor had a 20 In the second condition, subjects knew percent chance of requesting 7 points, they had an opportunity to exchange in a 60 percent chance of choosing 8, and an alternative network and the only dif- a 20 percent chance of choosing 9, result- ference between their existing network ing in an average initial request of 8 and the alternative network would be points and an average final request of 7 the people with whom they interacted. points.7 Subjects who accepted a final request of 7 received 5 points. Programming of Simulated Actors To create this 5/7 split in reciprocal exchange conditions, the program manip- We modified a program used by Molm ulated the frequency with which the et al. (2006) for a previous social exchange experiment to create simulated actors. 7There was a 10 percent chance of the high Modifications ensured simulated actors power simulated actor reaching an agreement behaved in realistic ways, appropriate to with another simulated exchange partner. In their structural positions, and were addition to enhancing realism, including this fea- informed by behavioral data from previous ture helped make the negotiated exchanges more similar to the reciprocal exchanges, where high exchange experiments using only human power simulated actors chose not to give to lower subjects (e.g., Molm et al. 2000). power participants.

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 124 Social Psychology Quarterly 79(2) simulated actor reciprocated the subject’s resulted in a scale with an alpha reliability giving. Specifically, we adjusted the prob- of .88. A similar scale for the low value ability of the simulated actor giving after relation had an alpha reliability of .86. receiving and the probability of the simu- Psychological group formation. After lated actor giving after not receiving. The subjects were made aware of the alterna- probability of giving after receiving was tive network, they were asked to answer set at .68, and the probability of giving three questions designed to measure psy- after not receiving was set at .03. This chological group formation. The first item created the desired ratio. had subjects indicate whether they felt Interactions between the participant very unattached/very attached to the and the other simulated actor were three other participants with whom they designed to create, on average, equal had been interacting (Bargozzi and Lee value exchanges. In negotiated exchange, 2002). The second item asked subjects to this was achieved by having the simu- describe how much obligation (very lit- lated actor request 2 to 4 points in the tle/very much) they felt toward these first round, lower this request by 1 in other participants (Bargozzi and Lee the second round, and then repeat this 2002), and the third had subjects indicate request for all subsequent rounds. On the extent to which they felt a sense of average, the final request was 2.8 A mod- belonging with this set of participants ified tit-for-tat strategy created the (Bargozzi and Lee 2002; Thye et al. desired ratio in reciprocal exchange. 2011). All items ranged in value from 1 Here, the probability of the simulated to 7, with higher scores indicating greater actor giving after receiving was .85, and levels of psychological group formation. the probability of the simulated actor giv- The alpha reliability score for these three ing after not receiving was .15. items was .84. A factor analysis with var- imax rotation (not shown) of all of the Measures items for the personal attraction meas- Personal attraction. We measure per- ures and the psychological group forma- sonal attraction with four items designed tion measure produced a three-factor solu- to evaluate how subjects assess each of tion with the psychological group formation their dyadic exchange relationships along items loading on a unique factor and the these dimensions: divided/united, self- personal attraction measures for the focal oriented/team-oriented, coming apart/ actor and the personal attraction measures coming together, and partners/adversaries for the other actor also loading on separate (Molm et al. 2012). These measures factors. We are therefore confident that ranged in value from 1 to 7, with higher although related to the personal attraction scores indicating greater attraction. measures, the measure of psychological From these items, a personal attraction group formation is distinct. measure was created for each participant Staying behavior. We measured staying for each exchange relationship. Averaging behavior by having participants choose to responses for the high value relation join an alternative exchange network or not. Those who chose to stay in their 8We added realism to the experiment by having this low power simulated actor occasionally existing network received a 1. Those reach agreements with another simulated actor who chose to leave received a score of 0. after the second round of negotiations. For each exchange opportunity, there was a 6 percent Exchange frequency. Higher frequen- chance of this occurring. cies of exchange generate greater

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 Should I Stay or Should I Go? 125

Table 1. Means and Standard Deviations for Measures by Experimental Condition (N = 64)

Reciprocal exchange Negotiated exchange Measures Low info High info Low info High info

Personal attractions for Focal actor 4.94 4.48 4.00 4.17 (1.04) (1.22) (1.56) (.70) Other actor 3.28 3.75 3.03 3.05 (1.21) (1.19) (1.33) (1.27) Psychological group formation 4.33 4.54 3.5 2.88 (1.32) (1.04) (1.22) (1.41) Staying behavior .5 .5 .19 0 Exchange frequency with focal actor .76 .67 .67 .70 (.12) (.15) (.15) (.14) N 16 16 16 16

Note: Standard deviations in parentheses. relational cohesion between pairs of importance as participants experienced exchange partners, and the structure of power differences through this relation- exchange can affect exchange frequency ship. Exchange frequencies ranged in (e.g., Lawler and Yoon 1993, 1996; Molm value from 0 to 1 for each exchange et al. 2000).9 Because the frequency of relation. exchange might vary across the two forms of exchange and the two power conditions, we include exchange frequency as Results a control. We measure the exchange fre- Table 1 reports descriptive statistics by quency at each phase of the experiment condition for the aforementioned meas- by dividing the number of agreements ures. In line with expectations, mean with (negotiated exchange) or acts of giv- scores for personal attraction, psychologi- ing to (reciprocal exchange) the high cal group formation, staying intentions, value simulated actor by the total number and actual staying behaviors are gener- 10 of exchange opportunities. Exchange ally higher for the reciprocal exchange with the high value partner is of utmost conditions. The table also reports the exchange frequencies between subjects 9 Our measure of exchange frequency had a rel- and their two exchange partners for the atively high correlation with earnings (r=.77). We therefore did not include earnings in the sta- entire experiment. Across the two forms tistical models we present. Doing so, however, of exchange, the exchange frequencies does not result in different conclusions. See between subjects and high power simu- online Appendix B available at spq.sagepub lated actors were similar to those in an .com/supplemental for these results. experimental study conducted with only 10While our measure is consistent with that used by Molm, Collett, and Schaefer (2006), an human subjects (Molm et al. 2012). alternative way of constructing the measure for Because these descriptive statistics are reciprocal exchange conditions would be to con- consistent with our expectations, we test sider how often the participants received points our hypotheses more formally using from their focal partners. Our models are stable a series of regression models. Our argu- regardless of how we operationalize exchange frequency. See online Appendix B available at ment hinges on the idea that the exit spq.sagepub.com/supplemental for these results. opportunity prompts individuals to reflect

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 126 Social Psychology Quarterly 79(2)

Table 2. Unstandardized Ordinary Least Squares Coefficients from the Regression of Psychological Group Formation on Independent Variables (N = 64)

Independent variable Model 1 Model 2 Model 3

Form of exchange (Rec. = 1; Neg. = 0) 1.26* .81* .81* (.32) (.32) (.32) Exchange frequency with focal partner 2.24 2.29 2.34 (1.13) (1.09) (1.09) Personal attraction with focal partner .50* .50* (.14) (.14) Personal attraction with other partner .29* .29* (.13) (.13) Information 2.22 (.29) Constant 3.35* .48 .61 (.80) (1.13) (1.15) R2 .21 .36 .37

Note: Standard errors are in parentheses. p \ .05 (two-tailed tests). back on their past exchange relations and with psychological group formation being that this reflection will result in disad- higher in reciprocal exchange conditions. vantaged actors being more likely to Model 2 adds the two personal attraction undergo psychological group formation measures. The significant main effects if those past exchanges were charac- for both measures show that personal terized by reciprocal, as opposed to nego- attraction affects psychological group for- tiated, exchanges. We also contend that mation, and Sobel-Goodman tests reveal personal attraction at the dyadic level that both attraction measures partially will mediate the relationship between mediate the relationship between the the form of exchange and psychological form of exchange and psychological group group formation. formation, with personal attraction The regression analyses in Table 2 toward the high value partner mediating begin to test these claims. Model 1 about 32 percent of the total effect and regresses psychological group formation with personal attraction toward the equal on the form of exchange and the fre- value partner mediating about 21 percent quency of exchange between the subject of the total effect. These results provide and focal partner.11 Results show only support for Hypotheses 1, 2, and 3— the form of exchange has a significant following the exit opportunity, psycholog- effect on psychological group formation, ical group formation is stronger for disadvantaged actors located in a reciprocal exchange network and personal 11 Analyses run using the frequency of attractions for specific exchange partners exchange between the subject and other simulated actor as the control are generally consistent partially mediate this relationship. with those reported. The measure of exchange Next, we consider whether the amount frequency included in the reported results was of information actors have about the the average of the exchange frequency for the alternative network affects psychological first two exchange phases. Sensitivity analyses group formation. Model 3 in Table 2 reveal that only including exchanges that occurred during the first phase of exchange does reveals that increasing the amount of not modify results. information did not motivate greater

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 Should I Stay or Should I Go? 127

Table 3. Logistic Regression Coefficients Predicting Staying in One’s Social Network. The change in coefficients was derived using the KHB method. (N = 64)

Independent variable Model 1a Model 2 Change in b

Form of exchange (Rec. = 1; Neg. = 0) 4.36* 2.48* 1.88* (1.31) (1.00) (.90) Exchange frequency with focal partner 2.93 2.14 (3.18) (3.16) Personal attraction for focal actor 2.10 21.26* (.42) (.58) Personal attraction for other actor 2.32 21.00* (.41) (.48) Information 21.70 21.19 (.99) (.93) Constant 21.36 22.78 (3.49) (3.51) Psychological group formation 2.32* (.69)

Note: Standard errors are in parentheses. *p \ .05 (two-tailed test). aThe coefficients from this model differ from a standard logistic regression model because the residualized psychological group formation measure is included in the model, thereby making the coefficients in this model comparable to those in Model 2. We do not report the coefficient for the residual here. psychological group formation, as predicted But does psychological group forma- by Hypothesis 4. We, however, note that tion mediate the relationship between because we did not assess uncertainty per- the form of exchange and staying in ceptions directly, we do not know how par- one’s network as predicted by Hypothesis ticipants interpreted this information. 6? We answer this question using the Future research should more directly eval- KHB framework, which was created to uate the effects of uncertainty. compare coefficients of nested nonlinear Two questions remain: Does psycholog- models (Breen, Karlson, and Holm 2013; ical group formation affect whether indi- Karlson and Holm 2011). The KHB viduals ultimately choose to remain in framework is a multistep procedure that their existing exchange networks, and first regresses the mediator on the exoge- does it mediate the relationship between nous variables and retains the residual. the form of exchange and staying? This residual then is included in a reduced Hypothesis 5 states that because individ- model that does not contain the mediator, uals who view themselves as part of thereby ensuring that any difference a group come to favor that group over between the model with the mediator other groups, disadvantaged actors (i.e., the full model) and the reduced should be more likely to choose to remain model is attributable to the mediator. with their existing network the more they This process ensures that coefficients identify as a group member. A bivariate across models are measured on the same logistic regression analysis (not shown) scale and thus, comparable. yields support for this prediction. A one- Table 3 presents the results of an unit increase in psychological group for- application of the KHB method and the mation increases the odds of staying in mediation analysis. Including psychologi- the network by a factor of 4.60. cal group formation in the model results

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 128 Social Psychology Quarterly 79(2)

Table 4. Firth Logit Model Predicting Staying in One’s Network (N = 64)

Independent variable b (St. Error)

Form of exchange (Rec. = 1; Neg. = 0) 1.95* (.85) Exchange frequency with focal partner 2.02 (2.77) Personal attraction with focal partner 2.98 (.51) Personal attraction with other partner 2.80 (.42) Information 2.93 (.82) Psychological group formation 1.80* (.57) Constant 22.17 (3.08)

Note: Numbers in parentheses are standard errors. *p \ .05 (two-tailed tests). in a significant reduction in the main for dealing with small-sample bias (Firth effect of the form of exchange on the choice 1993). Results are in Table 4. They repro- to stay in one’s network, as predicted by duce the findings reported using the Hypothesis 6. In fact, psychological group KHB method, with the exception that formation explains 43 percent of the effect the personal attraction measures are no of the form of exchange on staying. This longer significant, as we would expect. finding holds even after controlling for Thus, we are confident in our conclusions power, information, and personal attraction. that disadvantaged actors in reciprocal We were surprised that including psy- exchange are more likely to stay in their chological group formation in the model networks and that this is in part because resulted in the coefficients for the two per- they are more likely to undergo psycho- sonal attraction measures becoming sig- logical group formation following the nificant and negative. Hogg (1992) con- exit opportunity.12 tends that personal attractions are often poor predictors of behaviors at the group EXPERIMENT 2 level, and as a result, we did not expect Our argument is that the opportunity to there to be a significant relationship, join an alternative network motivates much less a significant negative relation- psychological group formation for those ship. The small sample size and the rela- in reciprocal exchange networks but not tive lack of variation in the outcome for those in negotiated exchange networks the negotiated exchange conditions, how- and that this difference in group forma- ever, may have affected the stability of tion explains differences in staying our models and may account for this finding. We therefore ran another logistic 12 regression using the Firth method as We also ran a generalized structural equa- tion model to test the full model. The results are a sensitivity check. The Firth method, consistent with those we report here. See online which maximizes a penalized likelihood Appendix B available at spq.sagepub.com/supple function, is an efficient modeling strategy mental for these results.

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 Should I Stay or Should I Go? 129 behaviors. While results from Experiment negotiated exchange conditions (50), and 1 generally support this, the design of monetary values of points were adjusted Experiment 1 prevents us from ruling to ensure comparable behaviors received out the possibility that differences in psy- comparable earnings across the two forms chological group formation across the two of exchange. These steps were taken to forms of exchange simply reflect differen- ensure the two forms of exchange were ces in how people respond to reciprocal as similar as possible. Finally, the pro- and negotiated exchange, rather than grammed behaviors of the simulated how they respond following the exit actors were the same as in Experiment 1. opportunity. Experiment 2 attends to this possibility by examining whether Measures psychological group formation differs across the two forms of exchange absent Psychological group formation. We the exit opportunity. If our argument is relied on the same three items to measure correct, we would expect there to be no psychological group formation. Thus, our difference in psychological group forma- measure captures whether participants tion across these two forms of exchange. felt (1) attached to the other members of the network, (2) a sense of obligation to Design and Participants the other individuals, and (3) a feeling of belonging with the other individuals in Experiment 2 was a two-condition experi- the network. The alpha reliability for ment that manipulated the form of these three items was .77. exchange (reciprocal vs. negotiated). Thirty-six undergraduate students were Exchange frequency. The primary con- randomly assigned to each of these two trol variable for this analysis was the fre- conditions. We use data from 34 of these quency of exchange. Again, we measured because 2 students were suspicious about the exchange frequency by dividing the whether they were interacting with real number of agreements with (negotiated people. Each condition had 12 women exchange) or acts of giving to (reciprocal and 5 men. exchange) the high power simulated actor by the total number of exchange opportuni- Procedures ties. Exchange frequencies ranged in value from 0 to 1 for each exchange relation. Like Experiment 1, Experiment 2 involved participants engaging in a series of exchange opportunities with two exchange Results partners in what they believed was a fixed, To see if the form of exchange affected psy- four-person exchange network. The proce- chological group formation absent the exit dures were similar to those used in Exper- opportunity, we regressed psychological iment 1. The major difference was that group formation on the form of exchange after a set number of exchange opportu- controlling for exchange frequency. Table nities, Experiment 2 simply ended and 5 presents this analysis. Neither variable participants were asked to fill out a ques- was significant at the .05 level, which is tionnaire evaluating the extent to which consistent with previous research that they identified with the network as fails to show a difference in group identifi- a group. cation across these two forms of exchange Similar to Experiment 1, there were (Lawler et al. 2008; Molm et al. 2007). more exchange opportunities in reci- This finding strengthens our argument procal exchange conditions (100) than in that it is the exit opportunity that

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 130 Social Psychology Quarterly 79(2)

Table 5. Unstandardized Ordinary Least Squares Coefficients from the Regression of Psychological Group Formation on Form of Exchange and Exchange Frequency (N = 34)

Independent variable Model 1

Form of exchange (Rec. = 0; Neg. = 0) .74 (.44) Exchange frequency with focal partner 1.97 (1.36) Constant 1.55 (.96)

Note: Numbers in parentheses are standard errors. introduces into the situation a categorical exchange makes disadvantaged actors distinction that under the right conditions more likely to stay in exchange networks serves as the basis for disadvantaged than a history of negotiated exchange. It actors’ psychological group formation. shows that the greater personal attractions produced by reciprocal exchange prime disadvantaged actors such that DISCUSSION they are more likely to view themselves Much has been said about the bonding as group members following an exit power of the gift. Mauss (1925), for exam- opportunity. Thus, in addition to the ple, observed that gifts have both a mate- network-wide potential for exchange and rial and symbolic value that obligate gift the inequality of structural power (Lawler recipients to repay the favor. More et al. 2014; Thye et al. 2011), our research recently, social exchange theorists have points to the exit opportunity as another investigated how the different forms of factor that can motivate psychological exchange affect the solidarity that devel- group formation for disadvantaged actors ops between two actors (Kuwabara 2011; so long as it is primed by an appropriate Lawler et al. 2008; Molm et al. 2007). exchange history. They find that those forms of exchange This is important because group identi- that more closely resemble gift giving fication mediates a significant portion of result in stronger dyadic bonds than those the effect of exchange form on disadvan- that more closely reflect market-based taged actors’ staying behaviors. What exchanges (Kuwabara 2011). This is con- makes this finding different from past sequential given the prevalence of both research documenting how personal reciprocity and negotiation in the real attractions can explain an actor foregoing world (for accounts of reciprocal exchange potentially advantageous exchange rela- among contractors in the New York tionships with other partners in closed garment industry, see Uzzi 1997 or networks (Lawler and Yoon 1996) is that Moore 1978; for accounts of negotiated we make the network our unit of analysis exchanges over contracts in college coach- and thus turn to social rather than per- ing, see Greenberg 1991; for descriptions sonal attractions as the psychological of families that differ in the use of either basis for commitment. reciprocal or negotiated exchange, see Consequently, we speak to an apparent Collett and Avelis 2011). contradiction in the social exchange liter- Our research builds on this work by ature. While Emerson (1972) argues that examining whether a history of reciprocal power imbalances in exchange processes

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 Should I Stay or Should I Go? 131 should produce pressures toward struc- exchange than in pure negotiated tural change through coalition formation exchange (e.g., Molm et al. 2007, 2012), and network expansion, Blau (1964) our research reveals that this can result suggests that power imbalances could in disadvantaged actors foregoing oppor- become institutionalized and resistant to tunities to improve their situation. change. We account for these contradic- Exactly because reciprocity veils issues tory assertions by showing how the form of fairness by reducing conflict (Molm et of exchange can influence the likelihood al. 2003) and heightens the expressive of an alternative structural change value of exchange (Molm et al. 2007), it mechanism: exit. Unlike coalition forma- increases the likelihood of disadvantaged tion and network expansion, which actors adopting an attachment to the net- involve simply modifying linkages within work that can retard structural change. a network structure, exit involves sever- Thus, reciprocity as an integrating and ing ties with all members of a network stabilizing force may increase the likeli- and thus is another factor to consider hood of actors making choices that rein- when investigating how people respond force their disadvantage. to structural disadvantage. To know the generalizability of this Thus, we add to the literature on net- claim, additional research is necessary. work dynamics by uncovering how one’s We examine one particular power-imbal- exchange history informs how people anced exchange network. Research on respond psychologically to the exit oppor- negotiated exchange relations, however, tunity. Past theorizing and research on shows that variations in the structural the stability of exchange networks consid- properties of the exchange network can ers how uncertainty and the desire to affect group formation (Thye et al. 2011). accumulate resources informs the addi- It remains to be seen whether the findings tion and deletion of particular linkages generated here will hold in other power- between actors in a network (e.g., Dogan imbalanced network structures, and future et al. 2009; Emerson 1972; Kollock 1994; research should investigate this possibility. Willer and Willer 2000). Our research is Furthermore, our results indicate that different in that it examines situations psychological group formation only par- where disadvantaged people choose to tially mediates the relationship between extract themselves from whole networks form of exchange and network retention, rather than particular exchange relations, suggesting the possibility of other mecha- and it explores how the form of exchange nisms. One additional mechanism may be affects this choice through the psychologi- perceived risk. Previous scholarship finds cal process of group formation. Based on risk to be higher in reciprocal exchange our findings, we conclude that in addition than in negotiated exchange (Molm, to structural power, future investigations Schaefer, and Collett 2009). Given that into network dynamics should adopt people are loss averse (Kahneman and a more holistic account of the social struc- Tversky 1979), this may contribute to ture by considering the form of exchange whether people forego joining an alterna- and its effects on group formation. tive network. Yet another possibility is Our findings also cast a shadow on the that perceptions of hope for better future value of reciprocity. While Molm and col- exchange relations may vary across these leagues convincingly demonstrate that two forms of exchange and influence integrative bonds between exchange part- whether individuals stay. Schilke, Reim- ners are stronger in pure reciprocal ann, and Cook (2015) showed that low

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 132 Social Psychology Quarterly 79(2) power individuals tend to foster hope for the structurally disadvantaged and for the future. If the different forms of the lessening of structural inequalities. exchange influence what people hope for (escape from vs. improved relations with EDITOR’S NOTE network partners), this too may affect Both editors serve as editor-in-chief on all manu- the choice to stay. scripts. In this article, Richard T. Serpe served as Nevertheless, by addressing issues of the editor-in-chief because Jan E. Stets had a con- stability and inequality concurrently, flict of interest. this research yields valuable insights that have potentially broad implications. SUPPLEMENTAL MATERIAL For instance, our research suggests turn- Additional supporting information may be found over rates should be higher in organiza- at spq.sagepub.com/supplemental. tions characterized by a higher frequency of negotiated interaction, perhaps ACKNOWLEDGMENTS because this form of exchange yields a weaker psychological contract (Rous- We would like to thank Linda Molm, Joseph Gala- skiewicz, Kelly Bergstrand, David Melamed, and seau 1989). Similarly, our research may the members of the social psychology seminar at help explain the persistence or dissolution University of California, Riverside for their advice of marriages and families. Collett and on this paper. Earlier versions of this paper were Avelis (2011) reveal that exchange pat- presented at the 22nd Annual Group Processes Conference in Atlanta, Georgia, in 2010 and the terns in families tend to be dominated by 110th Annual Meeting of the American Sociologi- either reciprocity or negotiation. We there- cal Association in Chicago, Illinois in 2015. fore can surmise that low power family members will be more likely to sever ties FUNDING in families characterized by negotiation. While these hypotheses are mere specula- The author(s) disclosed receipt of the following financial support for the research, authorship, tions, they follow from the logic of our the- and/or publication of this article: This project ory and point to its general applicability. was funded by a National Science Foundation Thus, by explaining how micro-level Dissertation Improvement Grant (SES-1003281) processes can affect choices that ulti- awarded to Linda Molm and Scott V. Savage. mately affect the stability of macro-level REFERENCES structures, our research illustrates how theories of social exchange can inform Adler, Patricia A., and Peter Adler. 1995. sociological research more generally. It ‘‘Dynamics of Inclusion and Exclusion in Preadolescent Cliques.’’ Social Psychology also answers Molm and colleagues’ Quarterly 58(3):145–62. (2012) call to consider processes of rela- Backman, Carl W., and Paul F. Secord. 1962. tionship stability and change. With two ‘‘Liking, Selective Interaction, and Misper- experiments, we show that whether the ception in Congruent Interpersonal Rela- exit opportunity triggers psychological tions.’’ Sociometry 25(4):321–35. Bargozzi, Richard P., and Kyu-Hyun Lee. group formation for disadvantaged actors 2002. ‘‘Multiple Routes for Social Influence: depends on their exchange histories, with The Role of Compliance, Internalization, those in reciprocal exchange networks and Social Identity.’’ Social Psychology being more likely to undergo psychologi- Quarterly 65(3):226–47. cal group formation and thus, stay in Blau, Peter M. 1964. Exchange and Power in Social Life. New York: Wiley. their networks. Although this finding con- Breen, Richard, Kristian Bernt Karlson, and firms reciprocity as an integrating force, Anders Holm. 2013. ‘‘Total, Direct, and it also awakens us to its potential cost for Indirect Effects in Logit and Probit

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 Should I Stay or Should I Go? 133

Models.’’ Sociological Methods & Research Hogg, Michael A. 2012. ‘‘Uncertainty-Identity 42(2):164–91. Theory.’’ Pp. 62–80 in Handbooks of Theo- Campbell, Donald T., and Julian C. Stanley. ries of Social Psychology. Vol. 2, edited by 1963. Experimental and Quasi-Experimen- P. A. M. Van Lange, A. W. Kruglanski, tal Designs for Research. Chicago: Rand and E. T. Higgins. Thousand Oaks, CA: McNally Publishing Company. Sage Publications Ltd. Cheshire, Coye, Alexandra Gerbasi, and Karen Hogg, Michael A., and John C. Turner. 1985. S. Cook. 2010. ‘‘Trust and Transitions in ‘‘Interpersonal Attraction, Social Identifica- Modes of Exchange.’’ Social Psychology tion, and Psychological Group Formation.’’ Quarterly 73(2):176–95. European Journal of Social Psychology Collett, Jessica L., and Jade Avelis. 2011. 15(1):51–66. ‘‘Building a Life Together: Reciprocal and Homans, George C. 1974. Social Behavior: Its Negotiated Exchange in Fragile Families.’’ Elementary Forms. Rev. ed. New York: Pp. 227–54 in Advances in Group Processes. Harcourt Brace Jovanovich. Vol 28, edited by S. R. Thye and E. J. Law- Kahneman, Daniel, and Amos Tversky. 1979. ler. Bingley, UK: Emerald Group Publish- ‘‘Prospect Theory: An Analysis of Decision ing Limited. under Risk.’’ Econometrica 47(2):263–91. Cook, Karen S., and Mary R. Gillmore. 1984. Karlson, Kristian Bernt, and Anders Holm. ‘‘Power, Dependence, and Coalitions.’’ Pp. 2011. ‘‘Decomposing Primary and Second- 27-58 in Advances in Group Processes. Vol ary Effects: A New Decomposition Method.’’ 1, edited by E. J. Lawler. Greenwich, CT: Research in Social Stratification and JAI Press Inc. Mobility 29(2):221–37. Cook, Karen S., and Richard M. Emerson. Kmec, Julie A. 2007. ‘‘Ties that Bind? Race and 1978. ‘‘Power, Equity and Commitment in Networks in Job Turnover.’’ Social Prob- Exchange Networks.’’ American Sociologi- lems 54(4):483–503. cal Review 43(5):721–39. Kollock, Peter. 1994. ‘‘The Emergence of Cook, Karen S., Richard M. Emerson, Mary R. Exchange Structures: An Experimental Gillmore, and Toshio Yamagishi. 1983. Study of Uncertainty, Commitment, and ‘‘The Distribution of Power in Exchange Trust.’’ American Journal of Sociology Networks: Theory and Experimental 100(2):313–45. Results.’’ American Journal of Sociology Kurtzberg, Terri, and Victorica Husted Med- 89(2):275–305. vec. 1999. ‘‘Can We Negotiate and Still Be Dogan, Gonul, Macel A. L. M van Assen, Arn- Friends?’’ Negotiation Journal 15(4):355– out van de Rijt, and Vincent Buskens. 61. 2009. ‘‘The Stability of Exchange Net- Kuwabara, Ko. 2011. ‘‘Cohesion, Cooperation, works.’’ Social Networks 31(2):118–25. and the Value of Doing Things Together: Emerson, Richard M. 1972. ‘‘Exchange Theory, How Economic Exchange Creates Rela- Part II: Exchange Relations and Net- tional Bonds.’’ American Sociological works.’’ Pp. 58–87 in Sociological Theories Review 76(4):560–80. in Progress. Vol. 2, edited by J. Berger, M. Lawler, Edward J. 2001. ‘‘An Affect Theory of Zelditch, Jr., and B. Anderson. Boston: Social Exchange.’’ American Journal of Houghton-Mifflin. Sociology 107(2):321-52. Firth, David. 1993. ‘‘Bias Reduction of Maxi- Lawler, Edward J., Shane R. Thye, and Jeong- mum Likelihood Estimates.’’ Biometrika koo Yoon. 2008. ‘‘Social Exchange and 80(1):27–38. Micro Social Order.’’ American Sociological Greenberg, Martin J. 1991. ‘‘College Coaching Review 73(4):519–42. Contracts: A Practical Perspective.’’ Mar- Lawler, Edward J., Shane R. Thye, and Jeong- quette Sports Law Review 1(2):207–82. koo Yoon. 2014. ‘‘Emotions and Group Ties Hirschman, Albert O. 1970. Exit, Voice, and in Social Exchange.’’ Pp. 77–101 in Hand- Loyalty: Responses to Decline in Firms, book of the Sociology of Emotions: Volume Organizations, and States. Cambridge, II, edited by J. E. Stets and J. H. Turner. MA: Harvard University Press. Netherlands: Springer. Hogg, Michael A. 1992. The Social Psychology Lawler, Edward J., and Jeongkoo Yoon. 1993. of Group Cohesiveness: From Attraction to ‘‘Power and the Emergence of Commitment Social Identity. New York: New York Uni- Behavior in Negotiated Exchange.’’ Ameri- versity Press. can Sociological Review 58(4):465–81.

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 134 Social Psychology Quarterly 79(2)

Lawler, Edward J., and Jeongkoo Yoon. 1996. Molm, Linda D., Monica M. Whitham, and ‘‘Commitment in Exchange Relations: Test David Melamed. 2012. ‘‘Forms of Exchange of a Theory of Relational Cohesion.’’ Ameri- and Integrative Bonds: Effects of History can Sociological Review 61(1):89–108. and Embeddedness.’’ American Sociological Lawler, Edward J., Jeongkoo Yoon, and Shane Review 77(1):141–65. R. Thye. 2009. Social Commitments in Moore, Sally Falk. 1978. Law as Process: An a Depersonalized World. New York: Russell Anthropological Approach. Boston: Rout- Sage Foundation. ledge and Kegan Paul. Leik, Robert K. 1992. ‘‘New Directions for Net- Powell, Walter W. 1990. ‘‘Neither Market nor work Exchange Theory: Strategic Manipu- Hierarchy: Network Forms of Org- lation of Network Linkages.’’ Social Net- anization.’’ Pp. 295–336 in Research in works 14(3–4):309–23. Organizational Behavior, edited by B. Markovsky, Barry, John Skvoretz, David Wil- Staw and L. L. Cummings. Greenwich, ler, Michael J. Lovaglia, and Jeffrey Erger. CT: JAI Press. 1993. ‘‘The Seeds of Weak Power: An Exten- Rand, David G., Samuel Arbesman, and Nich- sion of Network Exchange Theory.’’ Ameri- olas A. Christakis. 2011. ‘‘Dynamic Social can Sociological Review 58(2):197–209. Networks Promote Cooperation in Experi- Mauss, Marcel. 1925. ‘‘Essai sur le Don: Forme ments with Humans.’’ Proceedings of the et Raison de l’Echange dans les Societes National Academy of Sciences 108(48): Archaiques.’’ Annee Sociologie 1:30–186. 19193–98. Molm, Linda D. 1994. ‘‘Dependence and Risk: Rousseau, Denise M. 1989. ‘‘Psychological and Transforming the Structure of Social Implied Contracts in Organizations.’’ Exchange.’’ Social Psychology Quarterly Employee Rights and Responsibilities Jour- 57(3):163–76. nal 2:121–39. Molm, Linda D., Jessica L. Collett, and David Rousseau, Denise M., and Snehal A. Tijori- R. Schaefer. 2006. ‘‘Conflict and Fairness wala. 1998. ‘‘Assessing Psychological in Social Exchange.’’ Social Forces Contracts: Issues, Alternatives, and Meas- 84(4):2331–52. ures.’’ Journal of Organizational Behavior Molm, Linda D., Jessica L. Collett, and David 19:679–95. R. Schaefer. 2007. ‘‘Building Solidarity Savage, Scott V., and Kelly J. Bergstrand. through Generalized Exchange: A Theory 2013. ‘‘Negotiating the Unknown: The of Reciprocity.’’ American Journal of Sociol- Role of Uncertainty in Social Exchange.’’ ogy 113(1):205–42. Sociology Compass 7(4):315–27. Molm, Linda D., and Karen S. Cook. 1995. Schilke, Oliver, Martin Reimann, and Karen S. ‘‘Social Exchange and Exchange Networks.’’ Cook. 2015. ‘‘Power Decreases Trust in Pp. 209–35 in Sociological Perspectives on Social Exchange.’’ Proceedings of the Social Psychology,editedbyK.S.Cook,G. National Academy of Sciences 112(42): A. Fine, and J. S. House. Boston: Allyn Bacon. 12950–55. Molm, Linda D., Gretchen Peterson, and Thibaut, John W., and Harold H. Kelley. 1959. Nobuyuki Takahashi. 1999. ‘‘Power in Nego- The Social Psychology of Groups. New tiated and Reciprocal Exchange.’’ American York: Wiley. Sociological Review 64(6):876–90. Thye, Shane R., Edward J. Lawler, and Jeong- Molm, Linda D., David R. Schaefer, and Jes- koo Yoon. 2011. ‘‘The Emergence of Embed- sica L. Collett. 2009. ‘‘Fragile and Resilient ded Relations and Group Formation in Net- Trust: Risk and Uncertainty in Negotiated works of Competition.’’ Social Psychology and Reciprocal Exchange.’’ Sociological Quarterly 74(4):387–413. Theory 27(1):1–32. Turner, John C., Penelope J. Oakes, S. Alexan- Molm, Linda D., Nobuyuki Takahashi, and der Haslam, and Craig McGarty. 1994. Gretchen Peterson. 2000. ‘‘Risk and Trust ‘‘Self and Collective: Cognition and Social in Social Exchange: An Experimental Test Context.’’ Personality and Social Psychol- of a Classical Proposition.’’ American Jour- ogy Bulletin 20(5):454–63. nal of Sociology 105(5):1396–427. Turner, John C., and Katherine J. Reynolds. Molm, Linda D., Nobuyuki Takahashi, and 2012. ‘‘Self-Categorization Theory.’’ Pp. Gretchen Peterson. 2003. ‘‘In the Eye of 399–417 in Handbooks of Theories of Social the Beholder: Procedural Justice in Social Psychology. Vol. 2, edited by P. A. M. Van Exchange.’’ American Sociological Review Lange, A. W. Kruglanski, and E. T. Higgins. 68(1):128–52. Thousand Oaks, CA: Sage Publications Ltd.

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016 Should I Stay or Should I Go? 135

Uzzi, Brian. 1997. ‘‘Social Structure and Com- investigating whether person identities petition in Interfirm Networks: The Para- can overcome structural pressures to dox of Embeddedness.’’ Administrative Sci- ence Quarterly 42(1):35–67. exploit others as well as how identities Willer, Robb, Francis J. Flynn, and Sonya Zak. can generate social inequality when struc- 2012. ‘‘Structure, Identity, and Solidarity: tural pressures are absent. He is also A Comparative Field Study of Generalized studying how status processes affect and Direct Exchange.’’ Administrative Sci- work group dynamics. His recent research ence Quarterly 57(1):119–55. appears in The Sociological Quarterly Willer, Robb, and David Willer. 2000. ‘‘Explor- ing Dynamic Networks: Hypotheses and (2016) and American Journal of Sociology Conjectures.’’ Social Networks 22(3): (forthcoming). 251–72. Zachary L. Sommer received his BA in BIOS sociology at Augsburg College, located in Minneapolis, Minnesota, in 2010 and his Scott V. Savage is an assistant professor MA in sociology at the University of Cali- in the Department of Sociology at the fornia, Riverside in 2012. His primary University of Houston. Currently, with research interests focus on the role of support from the National Science Foun- human emotions and identities in dation, he and his collaborators are exchange processes.

Downloaded from spq.sagepub.com at ASA - American Sociological Association on May 25, 2016