Identity-b ( i ) ased Intervention of Third Parties The Effects of Social Categorization during Mediation-Arbitration

The present research investigates the effects of social categorization on the intervention behavior of third parties who engage in the hybrid dispute resolution procedure of mediation-arbitration (Ross & Conlon, 2000). Specifically, it was predicted that an affiliation to a disputant leads third parties to favor the affiliated ingroup disputant over an unaffiliated outgroup disputant. Two studies support these predictions by demonstrating that unilaterally affiliated third parties engage in ingroup favoritism during arbitration, whereas non-affiliated third-parties (Study 1 & 2) and third parties affiliated to both disputants (Study 2) imposed balanced settlements. In addition to this, both studies identify third parties’ decision control, inherent to the two phases of mediation-arbitration as a relevant moderating variable for the emergence of this effect.


Introduction
Negotiations and third-party intervention are well-established methods that are frequently relied upon in order to solve social conflict (e.g., Pruitt & Carnevale, 1993).Any news medium can provide numerous examples of this which further indicates the prevalence of negotiations between disputants belonging to different groups (e.g., nations, companies, political parties, unions, etc.).While decades of negotiation research has focused on cognitive and motivational processes (cf. De Dreu & Carnevale, 2003), the social context of negotiations, as well as third-party intervention, has failed to attract extensive interest (Kramer & Messick, 1995;Thompson & Fox, 2001).In line with this notion, previous negotiation studies have predominantly focused on an interpersonal context, even though research on intergroup behavior has a long-standing tradition in the field of social psychology (see Bazerman, Curhan, Moore, & Valley, 2000;De Dreu & Carnevale, 2003, for revisions).Only few studies have realized intergroup negotiation contexts by means of either investigating negotiation teams (Thompson, Peterson, & Brodt, 1996), group representatives (Ben-Yoav & Pruitt, 1984), or intra-group negotiations (Kramer, Pommerenke, & Newton, 1993).However, it is important to note that these approaches examined parties to the negotiation whereas, to our knowledge, third parties assisting disputants throughout the conflict have not been systematically explored.
Previous research inspired by the social identity approach (e.g., Tajfel & Turner, 1979;Turner, 1985) has applied the minimal group paradigm (MGP; cf.Tajfel, Billig, Bundy, & Flament, 1971) to demonstrate that the mere categorization of individuals into arbitrary social groups can be sufficient to elicit ingroup favoritism (reviewed by Mummendey & Otten, 1998).More specifically, when allocating monetary resources, participants who perceive themselves as belonging to an (arbitrary) ingroup favor other members of their ingroup at the expense of outgroup members.
The present study contributes to existing research on intergroup negotiations, as well as to research on ingroup favoritism, by investigating whether social categorization affects the quality of third parties' proposals during the two phases of mediationarbitration (Ross & Conlon, 2000).The findings of previous research on intergroup negotiations and social discrimination (e.g., Mummendey & Otten, 1998) will be expanded in two ways.First, the present research aims to demonstrate that, in addition to intergroup disputants, social categorization may also influence the behavior of third parties by means of affecting the quality of third-party proposals.In line with the findings on ingroup favoritism, it is predicted that in the applied context of negotiations, third parties who share an ingroup identity with one disputant are in danger of biasing their proposals in favor of the ingroup member.
Second, the present research investigates a moderating variable for ingroup favoritism on behalf of intervening third parties.Thirdparty interventions distinguish between mediation and arbitration (Pruitt, 1981) which predominantly differ with respect to inherent decision control, i.e. the extent of which the third party can determine the negotiation outcome.The present research aims to demonstrate that the differential level of third parties' decision control in mediation and arbitration functions as a moderating variable for third parties' ingroup favoritism.Whereas mediators rely on the acceptance of both disputants in order to affect negotiation outcomes, negotiation parties relinquish their decision control to third party arbitrators, who in turn are free to impose any binding and irrevocable settlement upon the disputants.
In the following, theory and research on social categorization in (intergroup) negotiations are summarized.Subsequently, mediation-arbitration as a frequently applied form of third-party intervention is introduced.Finally, two studies are reported which investigate the effects of social categorization on the level of ingroup favoritism of third parties.

Social Categorization in Negotiations
Social Identity Theory (Tajfel & Turner, 1979) and Self-Categorization Theory (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987;Turner, Oakes, Haslam, & McGarty, 1994) have inspired significant contributions to research on intergroup behavior, in particular that of intergroup negotiations.The social identity approach provides a sound theoretical framework for the emergence of intergroup conflict.At the heart of this approach lies the idea that individuals inevitably engage in social categorization; the social world is divided into ingroups (groups to which one belongs) and outgroups (groups to which one does not belong).Based on social categorization processes, individuals in turn attain a unique social identity which is defined as -that part of an individual's self-concept which derives from his knowledge of his membership of a social group (or groups) together with the value and emotional significance attached to that membership‖ (Tajfel, 1978, p. 63).
Previous research has shown that membership to social groups, or precisely the individual's image of the social self (i.e., her or his social identity; Tajfel, 1978), can spur ingroup favoritism as well as intergroup competition (cf.Mummendey & Otten, 1998).Tajfel and Turner (1979) suggested that individuals with salient group memberships feel the need for a positive social identity and that such a positive ingroup distinctiveness can be achieved by means of discriminating in favor of the ingroup and at the expense of the outgroup (e.g., Tajfel et al., 1971).To summarize these assumptions vividly it could be said that in order for ‗me' to enhance my self-esteem, I will favor ‗us' (ingroup) over ‗them' (outgroup), so that in comparison ‗we' appear to be better than ‗they' are.Numerous studies using the MGP have supported these assumptions by showing that the mere categorization of individuals into arbitrary social groups suffices to trigger ingroup favoritism (Tajfel, 1978(Tajfel, , 1979;;Tajfel et al., 1971; reviewed by Mummendey & Otten, 1998).
Previous research investigating the effects of social categorization in negotiations has predominantly focused on disputants by investigating the behavior and outcomes of negotiation teams (e.g., Thompson et al., 1996) or group representatives which are held accountable by their constituents (e.g., Adams, 1976;Ben-Yoav & Pruitt, 1984).Numerous studies in this field demonstrated that intergroup negotiations are marked by a heightened level of competition (e.g., Benton & Druckman, 1973;Druckman, 1994Druckman, , 2004;;Klimoski, 1972;Morgan & Tindale, 2002).Consequently, the risk of impaired negotiation outcomes is increased (e.g., Ben-Yoav & Pruitt, 1984;Van Kleef et al., 2007).However, these studies did not investigate the effects of social categorization on third parties whose assistance is frequently relied upon in negotiations that feature increased levels of competition and which appear to have entered a stalemate (e.g., Pruitt & Carnevale, 1993).

Third-Party Intervention
The classic distinction in third-party intervention differentiates between mediation and arbitration (Pruitt, 1981). 1 The two forms of third-party intervention predominantly differ with respect to the inherent decision control, i.e. the extent to which the third party can determine the negotiation outcome (Thibaut & Walker, 1975).Throughout mediation, disputants retain the right to negotiate and develop a conflict solution; in the case of arbitration, negotiators relinquish their decision control to an arbitrator, do not negotiate and refrain from crafting their own solutions (Pruitt & Carnevale, 1993).Whereas arbitrators are empowered to dictate the settlement of a respective dispute, mediators lack this form of decision control (Carnevale & Pruitt, 1992;Kressel & Pruitt, 1989).
Given that both procedures imply unique strengths it has been advocated to sequentially combine mediation and arbitration and examine the effects of hybrid forms of third-party intervention (Ross & Conlon, 2000).In mediation-arbitration (Conlon, Moon, & Ng, 2002) a third party (a) attempts to facilitate agreements between disputants throughout a mediated negotiation and (b) arbitrates (i.e., dictates a binding conflict settlement) if the mediated negotiation fails to engender a voluntary agreement between the disputants. 2 Due to the specific advantages of both procedures, mediation-arbitration is a widespread technique in third-party intervention and regarded as a procedure of particular effectiveness in competitive intergroup negotiations (cf.Conlon et al., 2002;Loschelder & Trötschel, 2010).
Previous research on social categorization in negotiations has, to the authors' knowledge, not yet systematically varied social identities of third parties.Instead, it has predominantly investigated the behavior of disputants.With respect to the prevalence of mediationarbitration in dispute-resolution practice and the fact that it is considered a third-party intervention with particular applicability at competitive intergroup negotiations, it appears important to investigate the impact of social categorization during mediationarbitration.

The Present Research
The present research intends to demonstrate that a shared social identity can have a strong impact on third parties' intervention behavior and lead to ingroup favoritism in the hybrid third-party intervention of mediationarbitration (Ross & Conlon, 2000).Specifically, we tested the assumption that third parties who are affiliated to an (ingroup) disputant favor this party over an unaffiliated outgroup member in their arbitration proposals.During arbitration negotiators relinquish their decision control to the third party, who in turn dictates a binding and irrevocable settlement.In juxtaposition to arbitration, mediators do not possess such decision control and thus rely on the acceptance of disputants in order to impact outcomes.Based on this reasoning, it was predicted that affiliated third parties would refrain from ingroup favoritism during the mediation phase.In addition, it was predicted that third parties (a) without an affiliation to either disputant or (b) with an affiliation to both disputants would suggest unbiased intervention proposals during both mediation and arbitration.
Two studies have been conducted to investigate these predictions.Study 1 investigated the behavior of participants in the role of third parties during mediation and arbitration and systematically varied whether third parties were non-affiliated versus unilaterally affiliated to one disputant.Supplementing Study 1, Study 2 also tested the assumption that third-parties' affiliation to both disputants leads to equally balanced proposals as are put forward by non-affiliated third parties; irrespective of the level of decision control implied in the two phases of mediation-arbitration.

Study 1
In Study 1, participants assumed the role of third parties in a computer-mediated negotiation and were randomly assigned to one of two between-subjects conditions: (a) participants shared an ingroup identity with party A, while party B was declared to be an outgroup member (affiliated), or (b) participants functioned as third parties in a negotiation between two unaffiliated outgroup members (non-affiliated).In addition, Study 1 systematically varied third parties' decision control inherent in mediation-arbitration-participants first served as a mediator and finally conducted a binding arbitration.It was predicted that due to the lack of decision control third parties would not engage in ingroup favoritism in the mediation phase, irrespective of their affiliation (Hypothesis 1).In contrast, it was predicted that, due to their increased decision control, affiliated third parties would systematically favor the ingroup disputant during arbitration (i.e., show ingroup favoritism), whereas non-affiliated third parties would conduct a balanced arbitration (Hypothesis 2).

Methods and Procedure
Participants and design.Thirty-four students (21 females; age M = 21.4,SD = 2.33) from the University of Trier, Germany, participated in this negotiation study.Participants received 5 € for remuneration and were recruited through leaflets.The study followed a 2 (Third-Party Affiliation: affiliated vs. non-affiliated) x 2 (Decision Control: mediation vs. arbitration) design with repeated measures on the latter factor.
Negotiation task and experimental manipulations.Participants engaged in a computer-mediated negotiation with multiple distributive issues (cf.Murnighan, Babcock, Thompson, & Pillutla, 1999), in which all participants were assigned to the role of a third-party.Participants were told that the present study intended to investigate whether students' negotiation success is determined by their academic major.Hence, the upcoming negotiation would feature students from three German universities differing in their focus with respect to the offered majors: the University of Trier (Humanities & Social Sciences), the University of Kaiserslautern (Engineering & Applied Sciences), and the University of Mainz (Natural Sciences).Negotiations were allegedly hosted by a central server.In fact, the behavior of third parties alone was investigated while the behavior of disputants was computersimulated.Note that in line with previous research (e.g., Conlon, Carnevale, & Murnighan, 1994) all participants were assigned to the role of a third-party, whereas negotiators' behavior was standardized and thereby controlled for.
The negotiation task used an adapted version of the classic negotiation paradigm (cf.Pruitt & Lewis, 1975).In order to avoid potential demand effects (Kramer et al., 1993;Bottom & Pease, 1997) the negotiation task was not related to participants' social identities, i.e. much care was taken to avoid a fit between group identities and negotiation issues.Hence, third parties were informed that disputants were to enter simulated market negotiations and bargain over the distribution of 72 stocks.There were six different types of stock (type A to type F) with different quantities (i.e., 9 stocks of type A & D, 12 of type B & E, and 15 of type C & F).Instructions further explained that disputants would receive points for each stock and that it was the disputants' task to maximize their respective points.
Participants were informed that it was their task in the role of a third party to assist disputants while negotiating the distribution of the 72 stocks.Specifically, based on the mediation-arbitration procedure (Ross & Conlon, 2000), participants were first asked to propose a solution to the conflict as a mediator.Upon completion of negotiations, participants were asked to conduct a binding arbitration as disputants failed to come to a voluntary agreement.In line with previous research (e.g., De Dreu & Van Lange, 1995), disputants' behavior was simulated in that both disputants claimed all 72 stocks for themselves in their first offers, engaged in gradual concessions over the course of the negotiation and proposed final offers, in which they claimed a total of 48 stocks for themselves.Note that these competitive claims do not allow for an agreement (e.g., a compromise with 36 stocks for each party), thus rendering a binding arbitration on behalf of the participants inevitable.
All participants were students at the University of Trier with academic majors in Humanities or Social Sciences (e.g., Psychology, Economics, Philosophy).Participants' membership to their alma mater was made salient by means of (a) repeatedly presenting the crest of the University of Trier on their computer screens and (b) addressing participants as members of the University of Trier (cf.Haslam, Oakes, Reynolds, & Turner, 1999).The between-subjects factor of thirdparty affiliation systematically varied the university membership of negotiation party A. Whereas in the affiliated condition party A was declared to be a member of the University of Trier, in the non-affiliated condition party A was a student at the University of Kaiserslautern.Irrespective of the condition, party B was declared to be a student at the University of Mainz.To increase this manipulation of third-party affiliation the respective university crests were presented adjacent to disputants' offers and participants were solely referred to as members of their universities (Haslam et al., 1999).
Procedure.Per session, between six and ten participants were recruited.Upon arrival at the laboratory, participants were randomly assigned to an experimental condition and were individually taken to a cubicle with a networked computer.Participants received all subsequent instructions on their respective computer screens.They were given approximately 20 minutes to familiarize themselves with the negotiation task and to fill in a pre-dispute questionnaire.Three experimenters checked with participants to ensure that they understood the negotiation task, the computer program, and their respective roles.Prior to negotiations, participants were asked to log in at the central server where they would be randomly assigned to two disputants via their IP addresses.The log-in process was simulated and participants were matched with (simulated) disputants in line with the aforementioned manipulation of third-party affiliation.
Prior to the onset of negotiations, participants were asked to assume the role of a mediator and to propose a distribution for the 72 stocks on the bargaining table.Subsequently, negotiations commenced with disputants exchanging (simulated) offers.Participants were asked to pay close attention to negotiators' offers as they were to conduct a binding arbitration in the case that disputants failed to create a voluntary agreement.Disputants' final claims of 96 stocks (each party claimed 48 stocks) exceeded the maximally available 72 stocks and hence arbitration became inevitable.Participants were asked to assume the role of an arbitrator and to impose a binding settlement to the conflict.Specifically, for arbitration third parties were instructed to declare how many stocks they would assign to each party.Finally, participants were thoroughly debriefed and thanked for their participation.

Dependent variable and manipulation checks.
Third-party distribution proposals, ranging from 72 (all stocks assigned to party A) to -72 (all stocks assigned to party B) were assessed as the major dependent variable during both phases (i.e., mediation and arbitration).In addition, we assessed participants' identification with their alma mater (-I identify with the University of Trier‖, -I feel connected to the University of Trier‖, -I enjoy being a student at the University of Trier‖; Cronbach's α = .76;Simon, Trötschel, & Dähne, 2008;Stürmer & Simon, 2004).Items were accompanied by seven-point scales ranging from 1 (do not agree at all) to 7 (strongly agree).Finally, participants were asked to indicate the university membership of party A and party B, respectively.

Results
Analyses of the sex of participants did not reveal any significant main or interaction effects.For the purpose of simplicity, this variable is not included in the subsequent analyses.
Manipulation checks.An analysis of the degree of participants' identification with their alma mater revealed that this particular social identity of participants was successfully activated.As expected, participants in both conditions reported a high level of group identification (M = 5.75, SD = 1.10) that significantly differed from the scale mean, t(33) = 3.98, p < .001. 3 Analyses further revealed that participants in the affiliated condition (M = 5.53, SD = 1.23) did not differ from participants in the non-affiliated condition (M = 5.98, SD = 0.95) in their level of group identification, t(32) < 1.20, ns.With respect to the two items asking third parties to indicate the university membership of party A and party B, analyses showed that in line with our manipulations all participants correctly identified the two disputants as members of the University of Trier and Mainz (affiliated condition), as well as Kaiserslautern and Mainz (non-affiliated condition), respectively.

Disucssion
Study 1 was conducted to investigate the effects of social categorization on the behavior of third parties who engage in the hybrid dispute resolution procedure of mediationarbitration. Specifically, it was predicted that an affiliation with a disputant would lead to ingroup favoritism on behalf of third parties during binding arbitration (when their decision control is at its maximum) but not during mediation, which per definition implies a reduced level of decision control.The findings of Study 1 supported these predictions.First, third-party affiliation did not affect proposals during mediation (Hypothesis 1).Second, affiliated participants favored ingroup members during arbitration, while non-affiliated participants imposed more balanced conflict settlements (Hypothesis 2).
To shed further light on the effects of social categorization on the behavior of third parties in mediation-arbitration, we addressed two issues in Study 2. First, mediation proposals in Study 1 were assessed at the start of the mediation phase.It may be criticized that the absence of ingroup favoritism in this phase was rather due to this early assessment of proposals instead of the lack of decision control in the mediation phase.To corroborate our assumption that the anticipated transition from mediation to arbitration accounts for the emergence of ingroup favoritism, Study 2 assessed an additional third-party's proposals at the end of the mediation phase.Nonetheless, considering the lack of decision control in mediation it was predicted that third parties would refrain from ingroup favoritism (i.e., biased proposals) at both times; at the start and at the end of the mediation phase.Second, Study 1 did not realize a condition in which third parties are affiliated to both disputants.Study 2 addressed this shortcoming.Note, however, that an affiliation to both disputants is also not expected to produce ingroup favoritism as both disputants are members of the third parties' ingroup.

Study 2
Study 2 was conducted to further investigate the effects of social categorization on the behavior of third parties.Participants were randomly assigned to one of three betweensubjects conditions: (a) participants shared an ingroup identity with party A, whereas party B was declared to be an outgroup member (affiliated-to-one), (b) participants functioned as third parties in a negotiation between two unaffiliated outgroup members (non-affiliated), or (c) participants shared an ingroup identity with both party A and party B (affiliated-toboth). Study 2 varied third parties' decision control by asking participants to function as a mediator at the beginning and at the end of negotiations, as well as to conduct a final binding arbitration.We predicted that due to their lack of decision control, third parties would not engage in ingroup favoritism during mediation, irrespective of the time of mediation proposals (Hypothesis 3).In the case of arbitration, we predicted that third parties affiliated to one disputant would favor the ingroup member, whereas third parties without an affiliation as well as with an affiliation to both disputants would conduct a more balanced arbitration (Hypothesis 4) and thus would not differ from each other (Hypothesis 5).
Procedure.With minor exceptions, the procedure of Study 2 was identical to Study 1. Upon arrival at the laboratory, participants were randomly assigned to an experimental condition, taken to a cubicle with a networked computer and given approximately 20 minutes to familiarize themselves with the task and to fill out a questionnaire.Experimenters checked with participants to ensure that they understood the negotiation task, the computer program, and their respective roles.Participants then engaged in the simulated log-in process and were randomly assigned to two disputants in line with the aforementioned manipulations of third-party affiliation.
In contrast to Study 1, mediation proposals were not solely assessed at the beginning but also at the end of mediation.Again, as disputants' final claims of 96 stocks at the end of negotiations exceeded the maximum available 72 stocks, arbitration became inevitable, hence participants were asked to impose a binding settlement to the conflict by means of assigning stocks to each party.Upon completion of the arbitration, participants were debriefed and thanked.

Dependent variable and manipulation checks.
Third-party proposals were again assessed as the major dependent variable during both mediation and arbitration.To check the manipulations, the participants' identification with their alma mater was assessed using identical items as in Study 1 (Cronbach's α = .87)and participants were asked to indicate the university membership of party A and party B. Identification items were accompanied by seven-point scales ranging from 1 (do not agree at all) to 7 (strongly agree).

Results
Analyses on the sex of participants did not reveal any significant main or interaction effects.For the purpose of simplicity this variable is not included in the subsequent analyses.
Manipulation checks.Analysis on the degree of participants' identification with their alma mater revealed that this particular social identity of participants was successfully activated.As expected, participants in all three conditions reported a high level of group identification (M = 5.31, SD = 1.22) that significantly differed from the scale mean, t(47) = 7.45, p < .001.Analyses further revealed that participants in the affiliated-toone (M = 5.46, SD = 1.32), the non-affiliated (M = 5.48, SD = 1.07) and the affiliated-toboth condition (M = 5.00, SD = 1.28) did not differ from each other in their level of group identification, F(2, 45) < 1, ns.With respect to the two items asking third parties to indicate the university membership of party A and party B, analyses revealed that in line with our manipulations all participants correctly identified disputants as members of the University of Trier/Mainz (affiliated-to-one), Kaiserslautern/Mainz (non-affiliated), and Trier/Trier (affiliated-to-both), respectively.

Third-party proposals.
Third-party proposals were submitted to a 3 (Third-Party Affiliation) x 3 (Decision Control) ANOVA with the latter factor as a repeated-measures variable.The ANOVA showed a trend for the Third-Party Affiliation main effect, F(2, 45) = 2.09, p = .13,η 2 = .08,that was qualified by a Third-Party Affiliation by Decision Control interaction, F(2, 90) = 2.09, p = .11,η 2 = .08.As in Study 1, separate contrast analyses were again performed for mediation and arbitration proposals to test our hypotheses.In line with our predictions, third-party proposals did not differ during mediation, with mediators in the affiliated-to-one condition on average proposing equally balanced distributions (M = -0.44,SD = 1.36) as both non-affiliated mediators (M = -0.81,SD = 2.17) and mediators affiliated to both disputants (M = -1.00,SD = 4.03), t(45) < 1, p = .578(Hypothesis 3).Note that separate contrast analyses for the two mediation proposals corroborated this pattern of results, showing that participants did not differ across conditions at the beginning of mediation nor at the end, t(45) < 1, p = .732and t(45) < 1.12, p = .268(for beginning and end of mediation, respectively).

Discussion
The findings from Study 2 further corroborate the results from Study 1. it was shown that social categorization of third parties can lead affiliated arbitrators to favor an ingroup over an outgroup disputant (Hypothesis 4).However, in line with our predictions, this effect does not emerge during mediation (Hypothesis 3), in which third parties do not have explicit decision control and rely on the acceptance of mediators.In addition, Study 2 shows that third parties affiliated to both disputants do not differ in their mediation or arbitration proposals from unaffiliated third parties (Hypothesis 5).

General Discussion
The present research investigated identitybiased intervention behavior of third parties in the hybrid dispute-resolution procedure of mediation-arbitration.In line with our hypotheses, the findings demonstrate that during arbitration third parties engage in ingroup favoritism, that is, they favor a disputant over the respective counterpart when they share an ingroup identity with the former individual and perceive the latter as an unaffiliated counterpart.In contrast, entirely unaffiliated third parties (Study 1 & 2), as well as third parties affiliated to both disputants (Study 2), refrain from biased settlements during the irrevocable, binding arbitration.In addition, the findings support the assumption that this effect of social categorization on ingroup favoritism in third-party intervention is moderated by the level of decision control inherent to the respective phases of the dispute-resolution procedure.The aforementioned effect of third-party affiliation in arbitration did not emerge during mediation when third parties by definition have a lower level of decision control and thus rely on the acceptance of disputants in order to affect negotiation outcomes.

Hybrid Forms of Dispute Resolution
Previous research on third-party intervention has advocated sequentially combining the classic forms of mediation and arbitration as these hybrid forms of dispute resolution promise significant advantages over the singular approaches (e.g., Conlon et al., 2002;Ross & Conlon, 2000).It has been stated before that third parties were typically introduced in order to alleviate the competitiveness of negotiations that appear to enter a stalemate.Note that the competitiveness is particularly heightened in intergroup negotiations (e.g., Druckman, 2004) and that mediation-arbitration has been suggested as a dispute resolution procedure with particular applicability during competitive intergroup negotiations (e.g., Conlon et al., 2002).
The present findings demonstrate that when assigning third parties to an intergroup conflict, exceptional caution has to be applied.Not only may social categorization processes threaten the acceptance of proposals on behalf of disputants, a respective affiliation may also cause third parties to engage in preferential treatment of a disputant.This risk appears to be particularly increased during arbitration, when the influence of a third party is at its maximum.Future research should also investigate the hybrid dispute-resolution procedure of arbitration-mediation (Ross & Conlon, 2000) third parties commence with arbitration which only goes into effect in the case that disputants fail to reach an agreement in the subsequent mediation, and investigate whether the present findings generalize to this reversed form of third-party intervention.

Positive-Negative Asymmetry in Third-Party Intervention
The social identity approach predicts that individuals strive for positive social distinctiveness and thus engage in preferential treatment for the ingroup as compared to the outgroup.Interestingly, previous research has demonstrated that this effect of mere social categorization on ingroup favoritism does not emerge when the positive distinctiveness of the ingroup has to be established by means of allocating negative resources such as costs or burdens (cf.positive-negative asymmetry; Blanz, Mummendey, & Otten, 1997;Mummendey & Otten, 1998).When allocating negative resources, such as boring tasks or unpleasant noise, participants refrained from discriminating positively toward their ingroup at the expense of the outgroup (for a meta-analysis see Buhl, 1999).Note that in the present research participants assumed the role of third parties and were asked to distribute positive (i.e., stocks with profit points), as opposed to negative, resources.In this respect the task third parties engaged in during arbitration parallels the classic procedure in MGP research.Thus, it remains to be investigated whether the positive-negative asymmetry, which has been demonstrated for the mere categorization effect in MGP research, extends to the context of third-party intervention.More specifically, it should be tested whether third parties refrain from outgroup discrimination in a negotiation in which disputants bargain the distribution of negative resources (e.g., costs, burdens; Galinsky, Leonardelli, Okhuysen, & Mussweiler, 2005).

Limitations and Implications for Future Research
To our knowledge, the present research is the first to investigate the effects of social categorization on the behavior of intervening third parties during mediation-arbitration.Note that the present findings provide an initial insight into the potential impairments that can emerge from third parties' social identities.It should be emphasized, however, that in spite of the reliable findings in the two studies, we do not intend to suggest that third-party affiliation will inevitably lead to a preferential treatment during arbitration, nor that third parties' lack of decision control during mediation will automatically protect unaffiliated disputants from unequal treatment or exploitation.For instance, one might expect that identity-biased arbitration is reduced in the case that affiliated disputants violate established fairness norms or resist to yield whatsoever.With respect to mediation, future research should investigate whether third-party affiliation leads to ingroup favoritism on more subtle dimensions, such as the way third parties (differentially) treat disputants throughout mediation.
In addition, it remains an interesting research question to explore how mediation and arbitration proposals affect the perceptions and behavior of disputants depending on third parties' affiliation.Does an affiliation to the third party cause a disputant to expect ingroup favoritism and thus lead him/her to claim significantly larger pieces of the pie than in the case of no-affiliation or an affiliation to the opponent?Future research should closely investigate the consequences of the identity-based effect on third-party intervention and further examine the interactions between disputants and third parties in non-simulated negotiations.
Note that in the present research participants were asked to engage in the a conventional form of arbitration, during which third parties are asked to decide how each issue (i.e., type of stock in the present study) should be distributed among the disputants, irrespective of disputants' final offers.This conventional arbitration can be distinguished from finaloffer arbitration during which third parties declare the final proposal of one of the disputants as the binding settlement to the conflict (e.g., Grigsby & Bigoness, 1982).Future research should further examine the effects of social categorization on intervening third parties during final-offer arbitration in order to extend the present findings to this slightly altered procedure.
The present research induced a social affiliation of third parties by means of activating participants' social identity as members of their alma mater.This approach raises a few issues that should be addressed by future research.First, in contrast to previous research based on the SIA, we refrained from using minimal groups (e.g., Tajfel et al., 1971) but instead reverted to natural social identities (i.e., participants' university membership), hence reflecting many realworld negotiations, in which disputants and third parties are affiliated due to their respective group memberships.However, future research should additionally investigate social categorization processes in mediationarbitration using minimal group manipulations.Second, much care was taken to disentangle the negotiation task (i.e., the distribution of stocks) from participants' social identities in order to avoid potential demand effects (e.g., Kramer et al., 1993).Note, however, that the stock negotiations did not allow (simulated) disputants to make actual profit (i.e., no monetary incentives were implied).In light of these findings, future research will examine whether the reported effects are possibly intensified if (a) monetary gains/losses are at stake, (b) third parties are more emotionally involved, or (c) negotiations occur in a face-to-face setting as opposed to being computer-mediated.Third, future research may investigate whether the effects of identity-b(i)ased third parties are further moderated by the nature of the intervention.Idealistically, third parties are expected to assume a neutral position.In the case of contractual intervention by professionals (Pruitt & Carnevale, 1993) this might be less of an issue than during emergent intervention featuring third parties from the same organization/system (cf.Murnighan, 1986), who possibly strive to further and preserve their own interests (e.g., Zartman & Touval, 1985).

Conclusion
The present research intended to shed an initial light on the effects of social categorization on the behavior of third parties in intergroup negotiations.Specifically, we demonstrated that a shared ingroup identity can lead third parties to engage in ingroup favoritism, that is, they declare identity-biased arbitration settlements.In addition, the findings show that this effect of third parties' social identity is moderated by the level of decision control inherent to the two phases of mediation-arbitration.Thus, the present research contributes to better understanding of the social categorization processes of third parties and underlines the particular caution that has to be applied when assigning third parties to assist in intergroup negotiations.