Role Effects in Negotiation: The One‐Down Phenomenon

Role is a concept that underlies most studies of human behavior in negotiation. The typical experimental negotiation simulation asks participants to play such roles as buyer or seller, plaintiff or defense attorney, or labor or management representative (Morley and Stephenson 1977). Similarly, naturalistic studies focus on such roles as teacher and administrator contract bargainers, hostage takers and hostage negotiators, Palestinian and Israeli peace negotiators, and husbands and wives in divorce mediations (Donohue, Diez, and Hamilton 1984). To explore these roles, research has focused on a range of phenomena from psychological orientations — including framing effects and various predispositions to communication style — to linguistic aspects of strategies and tactics as bargainers interact (Lewicki, Saunders, and Minton 1999). Therefore, the contexts surrounding the research on roles and the phenomena that are explored are quite diverse.

Despite these many explorations of role, few studies have examined patterns of role‐related phenomena across contexts. However, a cursory review of research in this area reveals some prominent role effects. Our purpose in this article is to select several negotiation contexts that explore role effects and examine the kinds of differences that emerge in these contexts. We find that role is characterized by a “one‐down” phenomenon in which negotiators who face various cognitive and contextual parameters in an interaction are more likely to use aggressive behaviors as a way of altering these parameters to achieve their goals. We show how this phenomenon is evident in the findings of research in a diverse range of negotiation contexts, where typically (but not exclusively) the one‐down position has been associated with a particular role position in the context being examined. Finally, we tie our review of the literature into existing theoretical frameworks that might explain the differences that emerge and the psychological basis for the “one‐down effect.”

The classic concept of role comprises three general components that incorporate both cognitive and contextual dimensions (Rodham 2000). The cognitive dimension addresses the internal orientations that negotiators bring to the table. The first is expectations. Performing a role involves behaving in a manner consistent with a set of expectations driven both by the negotiator's own needs and the needs of the constituents. A union representative, for example, is accountable to both the union members and its administration, as well as to his or her own expectations of how an individual should act within that role. These could include the expectation that he or she must forcefully insist on union gains detrimental to the management's position.

Similarly, the “rules of exchange” predict that a hostage taker will show some degree of “negotiability” either through extending deadlines, lessening demands, or allowing certain hostages to be released (Wilson and Smith 2000). In this sense, the hostage taker's goal is to maximize personal reward while moving through a somewhat predetermined negotiation process. The extent to which negotiators fulfill their expectations is likely to reflect both individual differences and the strength of their context‐driven expectations. Some negotiators will be flexible and adopt a role in response to the demands of the audience; others, in contrast, will adhere to their own personal preconceptions about how one should act in the role (Koestner, Bernieri, and Zuckerman 1992). The influence of context‐driven expectations is likely to manifest itself in the form of differences in behavior across role, with some roles (e.g., union representative or management) being tied into a set of more rigid expectations about positions than those roles that are less defined (e.g., hostage taker).

The second general component of role is social role identity. If a negotiator has internalized his or her role, then that role becomes that individual's social identity within the situation in which the role is performed (King and King 1990), from which he or she develops an organized collection of beliefs and attitudes about himself or herself (Jackson and Smith 1999). The union negotiator internalizes a set of expectations that establishes how he or she wants to be viewed by other negotiators. The hostage negotiator, who commits to fast action in order to restore public order in his or her role as a law enforcement official, must confront this role identity and adopt an unhurried, more patient style of interaction to allow the crisis to defuse and play out safely. In both cases, the negotiator seeks a position that matches his or her sense of personal identity, with mismatches between position and identity typically provoking use of behaviors that reconcile the disparity. Indeed, expectations about social identity as well as the context of interaction affect every aspect of negotiation by influencing the encoding of new information, memory of old information, and inferences where information is missing (Fiske and Taylor 1991).

Interestingly, social identity can be expected to have much less impact on undergraduate students playing roles in simulations than on actual negotiators; role‐playing negotiators are less likely to have internalized the specific expectations associated with the role that lead to a well‐defined and often defended social identity. In contrast, individuals committed to an identity that defends internalized principles and has a long history of doing so will likely pursue different strategies in negotiating. Therefore, key distinctions between studies based on actual roles and studies based on experimental negotiation roles would also be expected given the importance of social identity in role.

The third general component of role is complementarity. A negotiator's sense of social identity is based on role expectations that are defined in relation to the other party. It makes no sense to have a union representative without a management representative or a hostage taker without a hostage negotiator. The role of each negotiator is thus shaped by the extent to which he or she feels dependent on the other for exchanging resources (Kelley and Thibaut 1978). This dependence may be understood in two ways. The first relates to the fact that a negotiator's ability to achieve his or her goals may or may not depend highly on the other party's actions (i.e., interdependence in the relationship). If parties highly value one another's resources, they must shape their ideas and proposals directly to the ongoing activities of the other. To achieve this, they may rely heavily on a set of expectations about each party's role and how to move through instances of proposals and counterproposals, questions and answers, and so on. The second part relates to the valence of this interdependence in terms of focusing on the differences or similarities between negotiator positions (Alper, Tjosvold, and Law 1998; Thompson and Hrebec 1996). By its very nature, negotiation is a mixed‐motive activity that requires each party to understand how his or her goals and roles differ from those of the other party. As this mixed‐motive bargaining leads to an increasing emphasis on differences, parties can become more competitive and more protective of their respective role identities. In the context of intensifying competition, the dependence among negotiators can reinforce whatever fixed‐sum bias the negotiators may have (Putnam and Roloff 1992), such that their need to preserve role identity trumps their ability to focus on substantive goals.

An intriguing example of this challenge comes from studies of the role of emotional dependence in negotiator relationships. In a recent study, Gerben van Kleef and his colleagues (2006) sought to learn whether low‐power negotiators concede more to an angry or happy opponent. In five studies using different methods, they found that participants who are highly dependent on their counterparts were much more susceptible to one another's emotional reactions. The negotiators with fewer alternatives to negotiation (i.e., lower in power) were much more concerned about their potential payoffs, and, as others have shown (Levine and Boster 2001), these negotiators are much more aggressive than higher‐power parties in selecting compliance‐gaining strategies. Thus, the dependence between parties appears to make lower‐powered individuals more susceptible to emotion in both making concessions and in their selection of more aggressive tactics.

The cognitive and contextual elements reviewed previously suggest that negotiators see themselves in more or less dominant positions relative to each other at particular times in a negotiation. This emerges when negotiators: (1) have highly inflexible role expectations, (2) maintain a strong commitment to a role identity, (3) face significant relative power discrepancies, (4) are involved in a highly interdependent relationship yielding high total power, and (5) typically face highly emotional partners. Our position is that when negotiators find themselves in a role exhibiting these features, they are more likely to use competitive tactics. We describe these less powerful negotiators — who have lower expectations, fewer bargaining options, and greater dependence on the other party — as being “one‐down” from the other party. In this position, their need to maintain social identity and act in accordance with expectations that are inherent in the one‐down role is predicted to lead to compensatory behaviors that defend personal position and seek to regain power. For example, negotiators who begin with a weaker bargaining position are likely to focus on gaining ground and making sure that they make the most out of the position. In contrast, negotiators who expect to succeed in the negotiations (one‐up) will have available a greater number of options that do not impinge on their overall position, thereby giving them greater freedom to risk their social identity with more open and conciliatory dialogue.

It is important to note that we are not exclusively linking the one‐down effect with a particular role in each of the contexts. On many occasions, role dynamics are reversed across negotiators or they change over time as the negotiation unfolds (Donohue and Hoobler 2002a). For example, there are situations in which husbands are more powerful than wives, union members are more powerful than management, and so on. Interestingly, however, the majority of the research we have reviewed has attributed a certain set of opportunities and role differences to a particular position in the negotiation context. While this attribution may reflect the majority of negotiation situations, it is important to remember that the one‐down effect emerges as the result of a particular role position in a particular negotiation, defined by contextual and personal issues in relation to the other (i.e., complementarity).

We have examined our expectations about the one‐down position by conducting a review of previous research from different negotiation contexts. In this article, we will first discuss the identity issues associated with various roles and then review the behaviors associated with these roles. To provide a context for our analysis, we begin with a description of the typical research methods used in negotiation and conflict.

Quantitative research in conflict and negotiation is typically conducted in both naturalistic and experimental settings. The typical naturalistic study uses transcripts of actual conflict or negotiation events derived from either videotapes or audiotapes recording the sessions. (These studies often fall under the category of “case studies” or “case analyses.”). A naturalistic study analyzes data that are gathered without any researcher intervention suggesting that any kind of survey of actual individuals involved in an event would not be a naturalistic study. In a hostage negotiation, the police typically record the sessions for later debriefing. In a divorce mediation, the disputing spouses seek to resolve custody and visitation disputes with the assistance of a neutral third party. These kinds of sessions are naturalistic because they are actual events that occur in natural settings. The only obtrusive feature present in these events is the recording equipment. We also note that some studies explore simulated negotiations in these contexts that are generated from training sessions involving professionals (Taylor and Donald 2004). These contexts are naturalistic in the sense that they are not collected by a researcher but are part of an organization's normal practices.

This research paradigm typically involves studies artificially constructed by researchers using student (and sometimes professional) subjects, acting out roles such as buyers or sellers, labor and management negotiators, or plaintiff and defense lawyers. The vast array of possible experimental manipulations includes varying the levels of trust between negotiators, different amounts of information about their settlement aspiration levels, or instructing them differently regarding concession strategies. In some cases, the dependent variables include language codes like those described for the naturalistic research paradigm to test hypotheses about the level of cooperation or competition, or the kinds of integrative (collaborative) or distributive (competitive) strategies bargainers might use in the negotiations. In other studies, the dependent variables focus on concession rates, offers, and demands. Some studies also explore perceptual dependent variables such as settlement expectations, interpersonal attraction, and outcome satisfaction. Still others examine the impact of the negotiation medium itself (e.g., face‐to‐face versus electronic), the impact of adding participants (e.g., multiparty negotiation versus dyadic), or the roles played by race, ethnicity, and gender.

Given this methodological background, the different negotiation contexts can now be reviewed for evidence of the role effect. Analyses of hostage negotiations focus on interaction and content analysis studies of communication and language use between hostage takers and hostage negotiators (e.g., Donohue and Roberto 1993; Taylor and Donald in press). As suggested previously, the person in the hostage taker role exhibits most of the cognitive and contextual characteristics of the one‐down effect. Hostage takers, whether criminally or politically motivated or psychologically disturbed, maintain fairly high expectations while being intensely committed to their roles because they are bargaining for their lives. This negotiation occurs in the context of limited power for the hostage taker because the first rule of hostage negotiation is that there is no negotiation without containment. This policy limits the hostage taker's power and creates lower relative power, but in the context of high interdependence. Finally, there is often a great deal of potential for emotional displays, but not usually on the part of the hostage negotiator who typically remains fairly calm.

Most of the research in this area appears to support this profile. Practitioners report that there are essentially three kinds of roles adopted by hostage takers (Cambria et al. 2002). Criminally motivated hostage takers are typically caught in the act of another crime and take hostages in an attempt to force the police to allow them to leave. These hostage takers are often seasoned criminals who have experience in playing the criminal role and interacting with the police. In contrast, two other types of hostage taker generally have little experience or understanding of the criminal role. The domestic hostage taker is generally a male who has taken himself (nearly all hostage takers are male) or a family member hostage in the context of some kind of domestic disturbance. The mentally disturbed hostage taker seeks to make some kind of delusional point about an issue, such as the need to be seen as Jesus Christ in the context of the Second Coming. All three types of hostage taker share a commitment to gaining recognition of their personal concerns and often a high level of distrust and dislike of the police. They typically perceive the police as controlling the progress of the interaction, trapping or blocking the hostage taker from achieving what it is he wishes to achieve. Thus, while the roles associated with this context vary across incidents, the identities that the individuals present and the issues with which they are concerned are more or less well defined.

Like the types of hostage takers, the research in this area is similarly diverse. For example, William Donohue and Anthony Roberto (1993) studied the communication strategies across ten actual hostage negotiation situations involving all three types of hostage takers. The results indicated that the police negotiators controlled the amount and pace of the discussions and the topics that were considered. The police were also more likely to use collaborative relational messages that show support and provide more information than were the hostage takers, while the hostage takers were significantly more likely to use aggressive strategies such as threats, demands, and language expressing negative affect.

Randall Rogan and Mitchell Hammer (1994) used another set of hostage negotiations to explore the use of face work in negotiations. Face work refers to the behaviors that negotiators use to project both their own competence and desire not to be controlled, as a means of intensifying their power. These researchers discovered that police negotiators relied heavily on efforts to restore the hostage taker's face (using language to portray the hostage taker in positive terms) while the hostage takers relied more on restoring their own face or making themselves look strong and in charge. Thus, the perpetrators were much more focused on defending their own identities than were the police negotiators, who showed little concern for personal identity and concentrated more on trying to support the perpetrators' identities.

In another study exploring role differences, William Donohue and Anthony Roberto (1996) compared actual hostage negotiations to simulated negotiations in terms of the kinds of integrative or distributive strategies that were used during the interactions. The data revealed that the hostage takers in authentic contexts used more distributive strategies, made more demands, and proposed fewer integrative (win–win) options than the hostage takers or police negotiators in simulated contexts. Although these simulated sessions were conducted by actual police negotiators, the effect was not as significant as that found with actual hostage takers, presumably because negotiators in simulated sessions have a much lower commitment to their identity and less dependence on the other for an acceptable outcome. The actual hostage takers were significantly more interested in protecting their identities than individuals in any other category of negotiation.

The possibility of role orientations influencing the negotiators' language choices in hostage negotiations is also evident in the results of research conducted by Paul Taylor (2002). This research showed that negotiators in hostage crises typically orient around avoidance (withdrawn), distributive (competitive), and integrative (cooperative) approaches to interaction. Taylor argued that these broad orientations emerge from negotiators' interpersonal predispositions and personal role expectations in the conflict, and showed that their characteristic behaviors formed a single dimension running from extreme crisis to normative problem solving. He demonstrated that negotiators who hold firm to specific orientations tend to use communication behaviors that express either one‐up (moves seeking power) or one‐down (moves conceding power) strategies. Consistent with earlier accounts (e.g., Wilson and Putnam 1990), Taylor found that police negotiators and hostage takers generally adopt consistent orientations to instrumental (task‐focused), relational (trust and liking‐focused), and identity (face‐focused) concerns at any one time, reasserting the central importance of roles to many aspects of negotiation dynamics.

This is one of the most common negotiation contexts and typically involves bargaining over labor contracts. Labor representatives seek improved conditions of employment for their members and typically have a well‐defined set of objectives and an expectation to make some gains. Management representatives typically try to create settlements that reduce the employer's labor costs and improve the employer's profitability. Within this context, labor negotiators push for concessions from the management negotiators who, in turn, seek to maintain their powerful position with minimal loss. Negotiators adopting these roles can suffer from misperceptions about one another's intentions and motivations, and these misperceptions can further accentuate the existing role differences.

Labor and management negotiators are highly interdependent but given the formalized nature of labor–management negotiations and the fact that they often use agreed‐upon methods and specific standards for making their decisions, the roles played by both parties are typically well defined, although these roles may change given third‐party interests or mediator intervention (Lewicki, Weiss, and Lewin 1992). As noted by Stephen Littlejohn (1995), “Labor and management negotiators may fight to achieve their objectives, but they at least understand and accept the game in which they are involved, a process of presentation, application of power, negotiation, and compromise.”

Early research in negotiation used many labor–management simulation games, as summarized by Ian Morley and Geoffrey Stephenson (1977). In the typical scenario, the labor negotiator seeks better employment conditions for the workforce from the management negotiator, who is in a more powerful position because he or she has a greater range of options or a better payoff schedule. In general, experimental studies of labor negotiations indicate that negotiators representing labor are more competitive and concede less frequently than negotiators representing management. Developing these early findings further, many studies in communication have focused on the kinds of competitive and cooperative behaviors that bargainers use during these simulations. Research has shown that labor negotiators typically make frequent use of hard‐bargaining behaviors such as argumentative expressions, personal attacks, and unrealistic demands (Folger, Poole, and Stutman 1993; Putnam and Wilson 1989). In contrast, management negotiators are more likely to use more integrative behaviors that communicate a willingness to make concessions, develop a trusting and friendly relationship, and express confidence in the other party's abilities (Olekalns and Smith 2000; Putnam and Jones 1982). Thus, as Steven Wilson (1992: 196) argued in his summary of bargaining research focused on labor–management sessions, there are “several studies that show that labor tends to use attacking, offensive bargaining tactics while management uses defensive tactics.”

Studies examining the dialogue of actual labor–management negotiations typically concur with the conclusions of simulation studies. For example, David Bednar and William Curington (1983) found that local union representatives issued relatively more dominant and controlling statements and less deferential messages than management negotiators. Moreover, union negotiators asserted relational priorities using relatively more persuasive behavior and strategic messages but less task‐oriented and affective behavior than management negotiators. These dynamics stayed fairly consistent over the duration of the negotiation. The differences between union and management negotiators were even more pronounced when the union's spokesman was an international representative, presumably because these negotiations had higher stakes attached to their outcome.

Linda Putnam and Tricia Jones (1982: 171) analyzed actual teacher‐union and school‐board negotiations from a case in which the teaching staff felt poorly treated and were striving to improve their employment conditions. They found that “management representatives relied on defensive tactics while labor negotiators specialized on offensive maneuvers; these strategies emerged in the interaction structure of negotiators, especially in their use of attack‐defend and offensive‐information giving patterns.” Similar patterns were evident in Linda Putnam, Steven Wilson, and Dudley Turner's (1990) examination of policy arguments in teacher bargaining. Again, the teacher‐union negotiators used more messages that threatened the integrity of the management's identity as a way of forcing conciliation and weakening management's bargaining position.

In a study that compared a set of actual labor negotiations to a set of simulated sessions using professional negotiators, William Donohue, Mary Diez, and Mark Hamilton (1984) found that labor negotiators were nearly twice as likely to use attacking strategies compared to management, while management negotiators were twice as likely to defend themselves in response to an attack. In addition, management was twice as likely to use integrative strategies aimed at sharing information and problem solving than were the labor negotiators. These results were consistent across both the simulated and naturalistic sessions.

More recently, Maria Koutsovoulou (2001) provided even clearer evidence of how roles are associated with relatively consistent modes of self‐presentation that are reinforced by the reactions they elicit from others. In examining fifteen French labor negotiations, Koutsovoulou found that messages indicating flexibility such as asking for suggestions and giving positive evaluations typically encouraged the expression of other flexible, solution‐oriented messages and the suggestion of new alternatives. In contrast, hostile messages (e.g., rejecting offers, negative affective intervention) and pressure tactics produced further disagreement and negative affect.

Through this reciprocal process, roles seem to take on substance and become increasingly defined over time. This, in turn, suggests that the one‐down phenomenon will become more pronounced over time as negotiators reinforce their positions and expectations about the other party's and their own personal orientation. Evidence indicating that integrative solutions are less likely to emerge during the later stages of the negotiation process would seem to be consistent with this idea (Mannix, Tinsley, and Bazerman 1995).

Perhaps the most commonly studied negotiation context is the one involving buyers and sellers (Neale, Huber, and Northcraft 1987). The behavior of negotiators in this context presents a fairly mixed picture of the one‐down effect, particularly in the experimental research focusing on role. Although buyers and sellers are interdependent, their expectations and identities are not completely compatible (Dwyer, Schurr, and Oh 1987). For example, sellers in corporate settings typically aim to establish longer‐term contracts for larger quantities because this maximizes production efficiency and minimizes the impact of their initial standing costs. In contrast, buyers prefer to form short‐term deals for a smaller number of goods because this allows them to adapt more quickly to changing consumer demands (Nauta, de Vries, and Wijngaard 2001). This is certainly how the roles are conceptualized in the experimental research. Because of these priorities, the role identities of buyers and sellers are often different. Success for the buyer is usually focused on the price at which the goods are obtained. The seller, however, has a large number of variables to consider in meeting the buyer's demands, including the relationship with the buyer, product distribution issues, product quality, and servicing concerns. Thus, while the seller's position is more complex, it also provides more opportunities to resist the buyer's efforts to reduce price by offering alternatives.

Interestingly, in comparison to other contexts, evidence to fulfill our criterion for a one‐down position is less easy to identify in buyer–seller studies. Both parties maintain high expectations, but neither party seems to be egocentrically involved in his or her role, largely because of the experimental setting. There are potential power discrepancies because the buyer may have more options than the seller, but this relationship varies and these options often provide the seller with opportunities to resist the buyer's efforts to lower price. In real markets, the buyer might respond to this by moving to a different seller, but this is not possible in the experimental context, thereby often leaving the buyer in the one‐down position. Finally, while these contexts can get emotional, one role is not inherently more inclined to become more emotionally involved than the other. Thus, the association of a negotiator to a one‐down position is less clear‐cut in this context, although behaviors associated with the one‐down position do emerge.

To illustrate, in the typical study (as reviewed Morley and Stephenson 1977), students play the roles of buyers and sellers in mock negotiations. The typical task involves negotiating over a number of consumer‐type items with each party's maximum profits deriving from different items, thereby giving the task significant integrative potential. Creating this integrative solution requires that parties share information about their profit potentials and agree to cooperate in shaping the agreement. Findings from these early studies found buyers showing more fixed‐sum, win–lose biases and the competitive behaviors associated with these biases. These findings also correspond with negotiators' perceptions of their position, with buyers having a greater tendency to see aggressive strategies as being more effective to gaining a resolution (Bazerman et al. 2000; Diekmann, Tenbrunsel, and Galinsky 2003).

More recent work has sought to identify situations that may modify the relational roles of negotiators and, consequently, their communication behavior. For example, Laura Drake (2001) examined the buyer–seller communications and the influence of fixed‐sum errors on the interaction. She discovered that buyers held greater fixed‐sum errors than sellers at the beginning of the negotiations, but that sellers had more errors at the end of the interaction. The magnitude of the buyer error at the beginning of the interaction, however, was twice as large as any seller errors throughout the negotiations. She further discovered that, for the buyers, there was a negative relationship between information exchange and fixed‐sum errors throughout the negotiations, suggesting that the large buyer fixed‐sum bias at the beginning of the interaction severely attenuated information exchange throughout.

In an exploration of communication exchanges in these same data, Deborah Cai, Steven Wilson, and Laura Drake (2000) confirmed that buyers were more likely to give directional information, provide threats, and use put‐downs to enhance their positions. Their results are exceptionally interesting because the subjects were graduate students who were mostly from Asia, the Middle East, Europe, and Africa. Although they all negotiated in English, it is interesting to note that these results, with respect to role, were very robust culturally.

The aggressive behavior of these buyers in these experimental tasks seems to provide evidence of the one‐down effect. The key element in this context is the tendency of buyers to maintain a more highly developed fixed‐sum bias. Identities, power differences, total power, and highly emotional exchanges are not part of the typical formula in the experimental setting. Thus, it is probably fair to conclude that the observed differences in behaviors may sometimes not be a product of the one‐down effect but rather a product of fixed‐sum bias.

Several studies provide interesting insights about roles associated with international conflicts. For example, in their review of simulated negotiations between the United States and the Soviet Union, Morley and Stephenson (1977) found that the Soviet negotiators were more hostile and conceded less than the U.S. negotiators. Similarly, in examining bilateral negotiations between Spain and the United States over military base rights, Daniel Druckman (1986) demonstrated that negotiators moved through cycles of using hard tactics, such as retractions, threats, and commitments, and soft tactics such as initiations, promises, and accommodations. Overall, the Spanish government, in defending their personal position, used three times as many hard tactics as the U.S. team. More importantly, the pattern of behavior used by one country was the inverse of that used by the other country, with U.S. soft behavior accompanied on the whole by Spanish hard behavior, and vice versa. This observed mirroring of behavior was substantiated by a strong negative correlation in the degree of competitiveness between the two counties, particularly during the later rounds of the negotiations. Thus, although the Spanish government assumed the one‐down position for the majority of the negotiations (e.g., they had fewer bargaining options), occasions did arise in which the balance of power shifted and the U.S. team resorted to using more aggressive strategies.

Deborah Goodwin (2001) makes clear the importance of expectations and perceptions on negotiators' actions in international military conflicts. Developing the arguments of others (George and Simons 1994), she argues that the effectiveness of negotiators' strategies depend upon the other party's perception of its power and its ability to follow through on threats. When negotiators have a clear situational powerbase (i.e., one‐up), then this opens up a range of possibilities and “breathing space in which to alter decisions and approaches if necessary” (Goodwin 2001: 76). However, dominance in the negotiation may also enable a negotiator to make ultimatums and coercive threats that are acted upon by the other party because of the credibility afforded by the powerbase. Whether negotiators use the one‐up position to create possibilities or force a resolution depends on the immediate risks and time pressure associated with the conflict situation.

Consistent with these findings is a series of studies by Donohue and Hoobler (2002a, 2002b) that analyzed the power and affiliation messages that Palestinian and Israeli negotiators displayed in their public rhetoric from six months before the Oslo I agreement to six months before the Oslo II agreement in 1995. As others have noted, the Palestine–Israeli conflicts involve strong differences in identity (Litvak‐Hirsch, Bar‐on, and Chaitin 2003) as well as historically routed expectations (Kelman 1999). Power messages contain language aimed at showing toughness, resolve, vilification of the other, etc. Affiliation messages focus on showing flexibility in a position, supporting the other's rights or views, etc.

In the months leading up to the Oslo I agreement, when affiliative language was fairly low on both sides, the Palestinians used more power‐laden language than the Israelis, consistent with the norm of a low‐power or opposition party versus a more powerful adversary. A somewhat different pattern is evident in the lead up to the Oslo II agreement. In these interactions, power intensity was relatively equal while affiliation intensity was substantially different, with the Israeli leaders using about three times the level of affiliation as the Palestinian leaders. Given their perceived dominance over the situation, the Israeli rhetoric displayed a trend toward more affiliation, possibly in an attempt to generate interaction at the expense of being in complete control over the interaction. More importantly, in both sets of interactions, the U.S. team became counterpoised along a single relational dimension (i.e., affiliation or power), with patterns of relative positioning on this dimension shaping the role differences across the negotiation process.

This context appears to support the one‐down effect. Specifically, both parties maintain high expectations, are very committed to their respective identities, and often exchange information in very emotionally charged contexts. It also seems clear that the Palestinians hold relatively lower power compared to that of the Israelis, given Israeli access to resources and their ability to be mobile and dominate militarily. Total power between the parties appears high, however, as parties are highly interdependent in their conflict. Thus, it seems reasonable to conclude that the one‐down effect is at play in this dispute context.

Divorce mediations also provide a fairly clear example of the one‐down effect. The common scenario involves a husband who seeks to change visitation arrangements to give him more time with his children. In this situation, husbands typically take an offensive role in which they forcefully make a case for more visitation time. In resisting this change, wives are likely to adopt a more defensive, conciliatory role. Consequently, we might consider the husband to be in the one‐down situation and expect him to use more aggressive tactics than the wife, who may perceive that her position affords her stability that allows for more problem‐oriented discussion.

Most of the studies of divorce mediation provide data on the different bargaining approaches used by husbands and wives. For example, in a sample of twenty pre‐ and post‐divorce custody and visitation disputes from California courts, William Donohue (1989) analyzed the kinds of issues discussed by husbands and wives (e.g., issues based on fact, interest, values, or relationships). The data revealed no substantial differences between husbands and wives in the frequency of issue topics, probably because the mediators controlled the discussions and sought equal participation in issue discussions. When the transcripts were divided into different time segments, however, the husbands were significantly more likely than wives to initiate relationship‐based topics related to such concerns as lack of trust or consideration for their rights during the second interaction period, which is typically just after the orientation period. The wives were much more focused on factual or positional issues during this second time period. These relationship‐focused messages often intensify emotions, decrease positional flexibility, and result in power‐struggle conflicts that mediators often have difficulty controlling.

Following up on this relational discrepancy, Donohue (1991) also analyzed the levels of “verbal immediacy” present in the interactions. Verbal immediacy captures subtle verbal expressions that parties use to create or diminish psychological distance from each other (Weiner and Mehrabian 1968). One measure of immediacy, the implicit‐direct measure, focuses on the power dimension of language. More direct, less implicit linguistic references assert more rights to control the flow and direction of the interaction. In these twenty divorce mediation transcripts, the husbands were significantly more direct than the wives, reflecting the husbands' interests in generating more power moves during the mediation. Perhaps some of these differences are a result of gender differences in communication strategies (Pruett 1989), but in this naturalistic setting, it is difficult to parse out these gender effects.

In her analyses of these same transcripts plus some additional ones from other states, Tricia Jones (1989) focused on the kinds of strategies that divorce mediators use to address these power discrepancies. She discovered that effective mediator communication focuses on overcoming role‐status differences. Effective intervention strategies included facilitating information sharing, using summarization behaviors that promote supportive communication, avoiding reciprocity in the disputants' distributive behavior, and stimulating reciprocal problem‐solving behaviors. In negotiations where mediators rather than disputants take control of the interaction, the couple is more likely to engage in productive problem solving and reach agreement, presumably because the mediation minimizes the impact of role differences and encourages disputants to recognize their interdependence on the issue of their children's well‐being.

Jones's (1989) findings are consistent with related research on the negotiation of conflict issues between married couples. In analyzing the interaction of couples discussing a conflict issue, Alan Sillars and his colleagues (2000) have found that husbands were significantly more likely to focus on self, use issue appraisal, and see themselves as the ones instigating constructive engagement. In contrast, wives focused dialogue on their partner and the situation globally, and perceived their partner as using significantly more avoidance, detachment, and confrontation. Of course, these findings and others derived in the divorce mediation context are potentially explained by differences in male and female communication style rather than role. Indeed, many of the differences observed between husband and wife behavior may be attributed to gender differences (Pruett 1989), suggesting that it is too soon to attribute differences in behavior solely to role effects and the one‐down phenomenon.

Across these six rather diverse negotiation contexts, some interesting trends evolve with respect to role issues. The most striking trend is that negotiators who see themselves as having fewer options in comparison to their opponent are more likely to resort to aggressive strategies as a way of seeking change in the relative control and positions of each party. In other words, the person in the role most likely to seek changes will typically utilize competitive, aggressive messages as a means of shifting power, even though that strategy can be least effective in reaching agreement (Levine and Boster 2001). In existing research, hostage takers, buyers, union negotiators, low‐power international groups, and husbands are often found in roles that are characterized by cognitive and contextual conditions such as inflexible expectations, strong commitment to role identity, or significant power discrepancies. When these conditions exist, negotiators use more aggressive messages to promote a more competitive and more power‐focused negotiation context. The emphasis of their dialogue is on defending a personal position by attacking the other party's social legitimacy and attempting to force the other into unnecessarily yielding on critical issues. In contrast, the hostage negotiators, sellers, management negotiators, higher‐power international negotiators, and wives are less bound by these cognitive or contextual parameters. They use a broader range of communication choices including more affiliative and interest‐focused messages. As these negotiators have more options available that do not impinge on their overall position, they perceive that they are able to risk their social identity with more open and conciliatory dialogue.

As a starting point for a theoretical explanation of this one‐down effect, recall that role may be conceived as having three components: expectation, social identity, and complementarity (Rodham 2000). Negotiators have a collection of experiences and expectations that shape how they perceive the interaction and what they consider to be the most appropriate strategies for reaching a resolution. In conflict negotiations, where the emphasis is on differences among the parties, these expectations can often lead to negative frames (Drake and Donohue 1996) and fixed‐sum biases (Drake 2001). Such dynamics become prominent in interaction because expectations are integrated into negotiators' social identities, which they work to defend by regulating each other's control over the development of the interaction (Bacharach and Lawler 1986). The limits of this regulation are determined, at least in part, by the complementarity or dependence of each negotiator on the other party to achieve his or her goals. Thus, deriving a detailed theoretical account of how expectations, social identity, and complementarity work together would seem to form a fairly powerful explanation of the one‐down role phenomenon.

Expectations played an important part in the roles negotiators take up in research conducted in each of the six reviewed contexts: analyses showed that those who began the negotiation perceiving themselves in a position “one‐down” from the other party displayed more aggressive behavior at some point in the negotiation. The sources of these expectations, however, differed across contexts. While expectations in the experimental simulations were built around differences in task structure, in many of the real‐world contexts, expectations also seem to have been shaped by past experiences with the other party. The expectations of hostage takers, for example, were likely derived from a history of negative interaction with the police (Donohue et al. 1991), while in divorce mediations, each disputant has a substantial set of expectations and perceptions of his or her ex‐partner's motives and intentions (Donohue 1991).

In some contexts, such as trade union negotiations and international disputes, negotiators might also be influenced by external constraints (e.g., accountability to shareholders) (Wilson 1992). This possibility was confirmed in a study by Igor Mosterd and Christel Rutte (2000), which showed that negotiators who believed that they were accountable to constituents were much more likely to use competitive strategies and more likely to reach an impasse than negotiators working on their own behalf. It appears that accountability triggers greater scrutiny on the negotiator's role and the need to appear “tough” and committed to the perceived best interests of the constituents. Ironically, this identity trigger is counterproductive because the accountability results in significantly more impasses, thereby increasing constituents' costs.

Studies that examine negotiators' perceptions of role in terms of the cognitive‐heuristic frames that they bring to the interaction also demonstrate the importance of expectations (see Putnam and Holmer 1992 for a review). For example, Margaret Neale and her colleagues (1987) found that the assignment of negotiator roles, such as seller or buyer, cue gain and loss frames. Negotiators playing the role of sellers tend to view an outcome as a potential gain (i.e., those in an offensive role) and typically make more concessions and engage in more information‐seeking behaviors than buyers who tend to hold a negative frame. Negative frames are linked to escalation of conflict, potential impasse, strikes, and third‐party intervention. Negotiators entering interactions with negative frames are somewhat more likely to engage in defensive face work and somewhat less likely to engage in face work protective of the other side than are people entering negotiations with cooperative frames.

As previously mentioned, expectations become embedded in negotiators' broader identities and negotiators often act throughout a negotiation to defend these identities and the power structure that surrounds them. The notion that power and expectations interact is evident in Mara Olekalns and Bernhard Frey's (1994) study of buyer–seller negotiations, in which framing was shown to be exaggerated by differences in power. High‐powered positive‐framed negotiators and high‐powered negative‐framed negotiators are significantly advantaged by an imbalanced negotiation market in their favor. Thus, to ensure an adequate resolution of a conflict, a negotiator must be careful to monitor the situation variables that control the balance of power in the interaction and ensure that the power balance does not shift too far toward the other party. The roles of power and control within negotiation are, of course, much studied and are important aspects of negotiation theory (Bacharach and Lawler 1986).

According to Samuel Bacharach and Edward J. Lawler (1986), the extent to which regulation of control becomes necessary in bargaining depends on three factors. First, there needs to be a high degree of concern and commitment to social and public identity. Negotiators must value their identities, integrate them into all aspects of their social life, and work to defend them often. Second, partners need to identify norms and values that sanction violence in the broader social context. Violence must be seen within the relevant social group or society as an acceptable and appropriate means of restoring one's identity given the circumstances. The corollary for our exploration of the one‐down effect in different negotiation contexts is that being competitive and power‐focused, or integrative and affiliation‐focused, must be viewed as normatively sanctioned strategies for identity management. Third, individuals need to frame situations that call for the use of violence or power strategies. That is, they must work to actively structure the situation so that power, threats, or violence are viewed as a legitimate response to that situation. According to Bacharach and Lawler, then, much of what determines behavior within a bargaining situation is related to identity and the legitimization of force to restore it.

Through their analysis of ten international crisis case studies, William Zartman and Johannes Aurik (1991) expanded the focus from the individual who promotes threats or violence within a negotiation to an examination of how both parties contribute to the struggle for control. They describe how the more powerful party will often incite the less powerful party to invoke the need for the regulation of control. They indicate that the use of applied power — typically by the person in the more powerful role — may be effective in gaining short‐term objectives, but may also be likely to incite the less powerful party to retaliate later in order to regain or regulate control. For example, the high‐power role (e.g., the seller, hostage negotiator, wife, defense attorney, etc.) may use subtle or overt strategies to move the interaction in a particular direction that may be initially effective, but these moves also run the risk of challenging the identity of the lower‐power party. Zartman and Aurik (1991: 178) wrote that, in the international context, “whenever they were effective, threats tighten the jaws of deadlock, closing off further escalations and checking attempts to break out of the stalemate. At the same time, they also make the possibilities of negotiation appear more attractive, and, more specifically, make the second best look good by comparison.” However, even if the higher‐power party provides inducements to settle, the lower‐power party may perceive that his or her identity has been threatened in ways that make focusing on substantive issues challenging or impossible.

Therefore, is symmetrical power the solution? Frank Pfetsch and Alice Landau (2000) focus on this issue of power symmetry in conflict situations. They contend that symmetry is a necessary but not sufficient condition for successful negotiations. Properties of symmetry/asymmetry largely describe the relative potential power and strength of parties and thus power‐role status. Patterns of role behavior along these dimensions are seen in the process of negotiation — lower‐power parties will seek to negotiate on equal terms as the stronger party by cultivating nonsubstantive resources (e.g., future opportunities and supportive relationships) as a means of equalizing power and reestablishing identity. In this context, police negotiators often talk about the need to quickly restore the hostage taker's identity (e.g., through empathic listening) as a means of focusing on substantive issues (Donohue et al. 1991).

One problem associated with relying exclusively on this identity restoration construct to explain the role phenomenon observed in the six contexts cited is that the power effects were observed in contexts that did not meet the criterion of being highly identity intensive. The buyer–seller studies were simulations performed by students in which students were unlikely to have formed strong attachments to these particular simulated identities. However, even here the phenomenon emerged. As we noted earlier, there were significant differences in the fixed‐sum biases of the buyers in these studies. The fixed‐sum bias assumes that the task does not have an integrative potential and must be successfully competed with hard bargaining. Perhaps those in the low‐power position are significantly more likely to see that their only chance to succeed is to move aggressively against the high‐power negotiators to intimidate them into relinquishing control.

The importance of the mixed‐motive nature of negotiations highlights a third critical moderator of the one‐down phenomenon, which is the mutual interdependence among the negotiators (Sarbin and Allen 1968). The relative lack of identity issues associated with simulations is compensated for by the high degree of interdependence that derives from the requirement that the participants develop an agreement with each other (rather than a different third party) within a short time frame. From this perspective, the determinant of one‐down behavior seems to be the extent to which negotiators can quickly reach complementary positions focusing on differences with respect to the issues; that is, when parties face off directly on a task in either a real‐world or experimental context, they quickly discover the characteristics of their roles and their complementary nature.

Experimental studies that systematically vary the potential degree of interdependence (Beersma and De Dreu 2002) support this finding. Over a number of studies, greater interdependence yields more competitive behavior more quickly. The effect of interdependence is also evident in real‐world negotiations where parties are forced to reach a resolution. For example, Donohue and Roberto (1993) found that affiliation varies little across the progress of a crisis negotiation but that interdependence shows fairly stable patterns of change as the crisis moves toward and away from a solution. Similarly, in divorce mediations, mediators put considerable effort into helping divorced couples see past their differences and recognize that they are highly interdependent in their desire to create an agreement that maximizes the family life of their children (Donohue 1991).

In another context, Stephen Garcia, John Darley, and Robert Robinson (2001) argue that individuals assuming the defensive role typically show the most exaggerated behavior. In comparing powerful prosecuting attorneys (offensive role) with powerless defense attorneys (defensive role), Garcia and his colleagues found that negotiators in defensive roles were more prone to showing exaggerated perceptions, implying that perhaps in some cases, the defensive–offensive role of a negotiator is more predictive of exaggerated perceptions than the real distribution of power. If a negotiator's perceptions readily correspond to his or her behavior, then frequent use of aggressive distributive behavior might arise from the negotiator's perception that he or she is in a defensive position in relation to the other party, rather than simply less powerful. In other words, what negotiators' expect their role will require is closely related to the perceived power differences that emerge as the negotiation unfolds.

Dacher Keltner and Robert Robinson (1997) acknowledged this offensive–defensive dynamic in their study of traditionalists (powerful group) and revisionists (less powerful group) in the Western canon debate, a dispute about the content of Western literacy in which traditionalists argued against revisionists' desires to incorporate more works by women and minorities. In examining possible role differences through prospect theory (Kahneman and Tversky 1984), Keltner and Robinson (1997: 1068) conceded that traditionalists could “perceive the conflict as more costly, the demands and positions of their opposition more antithetical to their own, and the conflict more extreme.” They suggest that revisionists' (the less powerful challenging group) perceptions of traditionalists' bookpreferences and conviction were particularly exaggerated, leading them to use more aggressive behavior than the traditionalists as a way of rectifying the perceived imbalance.

These accounts raise the important question of whether or not there exist, in the review of the five contexts, examples of role dynamics that reverse the one‐down phenomenon. Changes in the power structure of the interaction could potentially create counterexamples of the one‐down effect. Reversals in the one‐down effect might arguably be associated with esoteric situations, such as a glutted market (Olekalns 1991) or such extreme kinds of interdependence as one might find in military situations (Goodwin 2001). Consistent with this idea, our previous study of terrorist negotiations (Donohue and Taylor 2003) found that interactions involving excessive demands by the terrorists typically led to an increase in affiliation between the parties and to better outcomes. Although this contrasts with expectations from other forms of negotiation in which making excessive demands is considered an aggressive strategy, authorities in intense conflict situations may conceivably view any form of dialogue as cooperative and a helpful inroad to resolving the incident. Thus, reversals of the one‐down phenomenon might be considered exceptions to the main effect, brought about by very specific sets of expectations and relationships among the two parties.

Reversals in the one‐down phenomenon might also be explained as part of the longer‐term oscillation between integrative and distributive behavior use reported by Daniel Druckman (1986). Evidence suggests that instances of overly positive or accommodating behavior by the one‐down party can reverse the one‐down phenomenon, prompting bargainers with positive frames (i.e., the one‐up party) to take advantage and engage in more risk seeking, more disagreement, and less concession making than the one‐down negotiator. In other words, when negotiators in powerful offensive roles perceive significant benefits to using their power, they may well use their advantage. This prompts an aggressive counterresponse from the low‐powered defender, which in turn forces the offensive negotiator to adopt more cooperative strategies in order to avoid a crisis or breakdown of the interaction. This pattern of interdependence may presumably carry on indefinitely.

How do these three perspectives merge into a coherent explanation for the one‐down phenomenon? The research indicates that we probably need at least two separate explanations: one to explain the phenomenon in students participating in simulated negotiations and another to explain its impact on seasoned practitioners entering an actual negotiation event. For students seeing a task for the first time and being assigned a role, it is likely that they first examine the task and then their roles within that task. If that task seems to be fairly unidimensional and the item for sale is of little value, then the student is likely to generate a fixed‐sum bias in the context of a negative frame. The need to manage an identity that is not salient or public is probably a low priority for these individuals.

For individuals who have more at stake and who must sustain their role identities for their sense of self, such as hostage negotiators, union negotiators, or spouses, the process is probably reversed. In this situation, these individuals must maintain their public identities and this will be relevant to their evaluation of the task. When these individuals perceive their position as being one‐down, they are more likely to enter the conflict with a negative frame and a focus on addressing the balance of power, particularly if affect is negative and the task is defined around a single issue as the Mosterd and Rutte (2000) study suggests. In contrast, negotiators who enter the same context with a one‐up position will have the same concerns about identity, but their available options and comfortable power position will encourage them to adopt a positive frame and an orientation toward identifying integrative solutions. The one‐down negotiator's focus is on ensuring that he or she is not “walked over.” The one‐up negotiator's task is one of reaching an acceptable solution in a fair and reasonable way.

This framework would predict that role differences would emerge depending on variables from two perspectives. From the individual perspective, differences would depend on the level of commitment to the role identity, the complexity of the task, and the extent to which the fixed‐sum and negative bias enter the dispute. From the dyadic perspective, differences would depend on the level of affect exchanged between the parties, and the extent to which the one‐up person flaunts his or her position rather than using it positively to open up integrative possibilities. When the one‐up party exercises his or her power and negative feelings dominate, then the need to control role identities for the low‐power party emerges, the task becomes artificially simplified, and the low‐power party assumes a fixed‐sum bias that focuses on avoiding loss.

Role effects in negotiation deserve greater attention. There seems to be compelling evidence that “one‐down” roles in conflict consistently produce more competitive strategies and more negative affect in negotiation contexts. The theoretical foundations of this phenomenon appear to center on role identities and biases at the individual level, and the exercise of power and negative affect at the dyadic level. Clearly, more studies need to be conducted and reviewed, and a tighter theoretical framework must be developed to better understand this phenomenon. Specifically, how strong is the phenomenon and can it be readily manipulated as some studies suggest? Research is underway to address these issues.