The Blackwell Companion to Sociology (96 page)

interpersonal interactions: the exchange of smiles, favors, etc. When I do some-

thing helpful for you, I often do not obtain explicit agreement that you will

return the kindness at a later time (that would be negotiated exchange). Any

benefit that I receive results from the individual action of the other person, and the extent of that benefit (if any) is unknown at the time that I make my decision about whether to act to help the other. This doesn't mean, of course, that

reciprocal exchange is not contingent. Obviously, if no behavior resulting in

value to me occurs, I am unlikely to continue sending rewarding behaviors

indefinitely. For the first 15 or 20 years, the research on negotiated and reciprocal exchange proceeded in parallel. Molm (2000a) has recently reported experi-

ments in which negotiated and reciprocal exchange are compared directly. She

found that power use was lower in reciprocal exchange than in negotiated

exchange. She also found that powerful actors in reciprocal exchanges were

more likely to opt for patterns of exchange that produced predictable, steady

exchange with their disadvantaged partners, rather than adopting a strategy of

intermittent rewarding that would have maximized their outcomes in the long

run. In other words, reciprocal exchange tends to produce satisficing rather than maximizing patterns of interaction that produce more equality than does negotiated exchange.

Position in both negotiated and reciprocal exchange networks determines how

often an actor exchanges with various partners, how many valued goods an

actor accumulates, etc. Actors are more dependent when they are connected to

fewer potential exchange partners, especially if those partners have many other

good options for exchange. Dependent actors are excluded from exchange more

often, get smaller quantities of valued goods in negotiated exchanges, and

receive fewer benefits from their relationships. The important feature here is

that this result requires no particular conscious or strategic action on the part of actors in less dependent positions. To have power is to use it. Less dependent

(more powerful) actors will receive better offers (in negotiated exchange) or

receive more frequent rewards from others (in reciprocal exchange) simply by

virtue of their position.

Much controversy in social network research has centered on formal measures

of structural power ± the advantage that a position derives from the availability of alternative exchange relations, the limits on the availability of their relations'

alternatives, and so on throughout a network. (See the June 1992 issue of the

journal Social Networks for a set of papers highlighting this topic.) Some

approaches have used network concepts like centrality or reachability, while

others use more formal graph analytic approaches. Other scholars have

challenged the usefulness of these purely structural descriptions of power by

410

Lynn Smith-Lovin

combining the influence of the network's structure and the actor's behavior

pattern on exchange outcomes (Molm, 1990; Freidkin, 1992). Actors who create

strong, contingent reinforcement patterns for their interaction partners achieve better outcomes than those who use inconsistent patterns. The effects of these

patterns are especially important for those who occupy relatively disadvantaged

network positions.

Because of Emerson's influence, much exchange research in the past 20 years

has focused on the voluntary exchange of positively valued things. The more

traditional (and colloquial) sense of power as coercive, negative control of

another's behavior has received much less attention. Most theorists who did

treat both reward-based power dependence and coercive, punitive power expli-

citly dealt with the two types of power as entailing different processes. Molm

(1997), however, incorporated both rewards and punishments within the power

dependence framework. She argued that you can be dependent on someone else

both as a source of valued things and to avoid things that you don't want to

happen. In experiments, this punishment power often is operationalized as the

ability to take points (worth money) away from another actor. This format

allowed Molm to investigate how reward power and punishment power worked

together to create outcomes in an exchange network.

Molm showed that people don't use punishment power very often, because it

is risky and people are generally risk-averse. They experience losses more sharply than gains. If you punish someone who has something that you want, that

person will often retaliate by withholding rewards or punishing you back, rather than doing what you want him or her to. So the only actors who use punishment

strategies very often are the people who don't have much to lose: coercion is the tool of people who are disadvantaged on reward power and who are imbedded

in highly imbalanced relationships (Molm, 1997, pp. 270±1). They have less to

lose and more to gain by using punishment. Punishment can be very effective,

however, if used with an effective behavioral strategy. As Molm (1997, p. 268)

puts it, ``successful coercion requires diligence in monitoring another's behavior, skill in applying punishment contingently, and the willingness to accept short-term losses (including normative censure) in return for uncertain long-term

gains.''

In addition to the new work by Molm comparing negotiated and reciprocal

exchange, new threads in the exchange literature highlight the structure itself as a dependent variable. More studies on coalition formation, network expansion,

and other forms of network transformation are making the relationship even

more central to this area. Some of this research explores how affective response as a social exchange outcome can promote cohesion and commitment in

exchange relationships; this work is reviewed below in an examination of the

convergence of instrumental and affective research interests. Other new research questions shift the focus from negatively connected, competitive exchanges to

positively connected networks in which resources gained through exchange with

one actor actually make exchange with another partner more likely. Since most

production systems have this character ± supplies need to be obtained from

multiple sources before a good can be produced ± these positively connected

Social Psychology

411

networks are very important for our understanding of social systems. There is

also a lively literature on how rational, goal-oriented actors can solve social

dilemmas ± situations where cooperation is mutually advantageous, but the

highest-payoff individual strategy is not to cooperate (see review in chapter 12

of Cook et al., 1995).

Expectation States Theory

The other major research tradition spawned by the early exchange theorists has

ventured much farther from its roots. Influenced by Homans's early exchange

theory and by Bales's (1950) observations of behavior in small task groups,

Joseph Berger and his colleagues (Berger et al., 1974) tried to explain why a

large number of group behaviors, like talking, being spoken to, evaluating the

ideas of others, receiving positive evaluations from others, tended to occur

together in what they called a power and prestige order. They proposed an

expectation states theory of how these observable status hierarchies developed.

They argued that under certain conditions ± a collective task orientation where

all group members stand to gain from better performance on a group enterprise ±

inequalities in task-related behaviors develop out of group members' expecta-

tions about the value of their own and others' contributions to the group

task. Effectively, deference is granted to group members in exchange for the

recipient engaging in behaviors that produce rewards for the other group

members.

Performance expectations are the central concept in the theory. Group mem-

bers form performance expectations, then give or take action opportunities to

produce the best group outcome, given their beliefs about group members'

relative abilities to contribute. Despite the fact that expectations are formed by individuals, however, the interactional encounter is the unit of analysis in the theory. In expectation states theory, it is the comparison of expectations for two group members that allows the prediction of the behavior that they will engage

in vis-aÁ-vis one another. Like power, status is a relational concept rather than an individual characteristic.

Performance expectations can form from several sources. In a group of highly

similar people, behavioral cues early in the interaction can be crucial. People

who engage in positive task behaviors, seizing early action opportunities, gen-

erate high performance expectations. But much of the expectation states

research focused on how status characteristics that were valued in outside

society were used to differentiate group members. When group members differed

on some evaluated characteristic, Berger argued that this characteristic became

salient in forming performance expectations. Unless there was evidence to the

contrary, the group member with the more valued state of the status character-

istic would be presumed by group members to have higher competence at the

group task than someone with the less valued status.

Berger and his colleagues adopted the strategy of isolating status processes

from other things that could go on in small groups (like emotional dynamics or

identity processes). They used a standardized experimental setting in which a

412

Lynn Smith-Lovin

subject interacted with a simulated actor. Both the subject and the simulated

actor made judgments about an ambiguous stimulus, under conditions where

they were motivated to make accurate judgments. After the individual judgments

were reported back to the subject, he or she was given a chance to change his or her individual decision if the two disagreed. The proportion of ``stay'' responses

± sticking with one's own opinion in the face of disagreement from the simulated other ± was the key dependent variable. Independent variables typically were

characteristics like gender, race, ethnicity, personal attractiveness, or education, that are evaluated in our society and that were assigned to the simulated actor in order to produce status advantage or disadvantage for the subject. The important thing about this research tradition is that these status characteristics (for example, being male as opposed to female) are not seen as directly determining

task behavior. Instead, the theory predicts that gender will determine the power and prestige order only when it is salient (because it differentiates group members) and when it is linked to expectations about the task in particular ways. For example, men are generally presumed in our society to be more competent than

women, but this is an indirect link to a non-gendered task. If the task is male-

stereotypic, the link could be stronger; if it is female-stereotypic, the link could be reversed in direction (with women receiving higher performance expectations).

Later work explored how demeanor, rewards from other tasks, and formal

positions could influence expectations (and therefore group behavior). After the basic structure of the theory was well supported, researchers devoted many

studies to exploring how different types of information combined to form

performance expectations (see review in chapter 11 of Cook et al., 1995). A

related thread of research investigated how status structures became legitimated, and how this legitimacy helped to stabilize the inequalities that result.

Expectation states theory has been very successful in applied research as well.

A number of experiments demonstrate how initial status differences, imported

from the external societal structure, can be reduced or eliminated by employing

interventions that shape group members' expectations. Elizabeth Cohen (1982),

for example, showed how group tasks and strategic interventions could help

racial minorities to become more active in school settings. Others have demon-

strated how gender inequalities can be reduced or reversed (see Wagner et al.,

1986, for an example and summary of this work).

Recent research has moved back from the rigorously controlled standardized

experiment to the analysis of small group interaction. For example, Cathy

Johnson (1993) has explored how being assigned to a managerial or worker

role in a simulated work environment influenced conversational behaviors like

interruption or suggesting ideas. As a consequence of this move to the study of

open interaction, researchers are beginning to investigate how status-organizing processes interact with other dynamics in groups (see discussion below of status and emotion). Another major focus of recent research is a new attention to the

dynamics of group interaction. Skvoretz and Fararo (1986), for example,

attempted to specify how information imported from outside status structures

combined with information that unfolds as the group interacts to form a

Social Psychology

413

cumulating status structure. New event history methods that allow the analysis

of behavior as it unfolds over time have allowed researchers to examine these

dynamics empirically.

Another important development in this tradition, also based partially on