Mediation Research: A Current Review

Aspirin and mediation have a lot in common: they are well‐known and easily recognized; both have been used for centuries; numerous articles have been published about the two. Both work, but we do not know why. Currently, biochemists and pharmacologists are laboring diligently to better understand aspirin. In this article, we pursue the same goals for mediation as we review the empirical, conceptual/theoretical, and applied scholarly works from the past ten years. Previous reviews (e.g., Wall 1981; Kressel and Pruitt 1989; Wall and Lynn 1993; Wall, Stark, and Standifer 2001) have thoroughly described the scholarly literature on mediation through the year 2000 and have established the groundwork for this review. Our goal here, as in the past, is to delineate where and how the literature has improved our understanding of mediation, and we will also indicate areas in which the scholarship is somewhat deficient. For this review, we considered scholarship focused on the mediation process itself; we did not consider articles or books on such related topics as mediation training, mediation ethics, policies governing mediators (e.g. training and certification requirements), career advancement for mediators (e.g. building a practice), or the representation of mediation in media and popular culture.

The report is organized as follows. First, we present a mediation paradigm that delineates the environment in which mediation is employed, limns the mediation process, and describes the mediation outcomes. After presenting this paradigm, we utilize it to organize the recent mediation literature, and, subsequently, we present an evaluation of relevant scholarship over the last ten years.

Using systems theory, as well as field theory, we begin (Figure One) with the basic assumption that mediation operates in a context or environment, and the literature indicates that this context comprises four major segments: conflict type, country, culture, and mediation institutions. This context or environment affects the mediation process. Herein, there is a mediator and two or more disputants (in Figure One, one disputant is represented for simplicity of presentation) who interact to determine the current state of the mediation. Within this interaction, the mediator has goals (e.g., agreement) and compares them with the current state (e.g., nonagreement). If there is a discrepancy, the mediator engages in certain behaviors — which usually include interactions with one or both disputants — to modify the current state.

Each disputant follows a parallel pattern of behavior. He or she has goals (e.g., high payment from the opponent) and compares them with the current state (e.g., the opponent offers to make a low payment). Finding a discrepancy, the disputant engages in behaviors to modify/improve the current state (e.g., threatens to leave the mediation if the mediator does not obtain a better offer from the opponent), and typically these behaviors entail an interaction with the mediator. This interactive mediation process has outcomes for the disputants, the mediator, and third parties who are not at the table but are affected by the outcome (e.g., payments to children in a divorce mediation).

We think it worth noting that the definition of mediation has remained constant during the past decade. Even though it has been lengthened, shortened, and fine‐tuned, after all modifications, the definition remains essentially the same: mediation is assistance to two or more interacting parties by a third party who — at that time — has no power to prescribe agreements or outcomes (Kressel and Pruitt 1989; Wall, Stark, and Standifer 2001). Most definitions from the last ten years are consonant with this one, and some scholars have embellished it by comparing mediation with arbitration, stating that in mediation, the third party controls the process but not the outcomes, whereas in arbitration, the third party controls the outcome but not the process. (In a few pages, we will note that this contrast is somewhat suspect.)

As Figure One indicates, the mediation context (or environment) comprises the conflict type, the country in which the conflict is mediated, the culture of the disputants and the mediator, and the mediation institutions available to the mediator and the parties. We consider first the conflict type.

In the past decade, the application of mediation to various conflict arenas has expanded voluminously — an extension that can be considered the hallmark for the literature in this period. As Table One indicates, we now find that mediation is utilized not only in the standard labor‐management, commercial, international, and marital conflicts, but also in environmental as well as community, civil court, intraorganizational, peace‐keeping, civil war, land claim, criminal, and child rights disputes.

These conflict types can be cleaved and categorized in multiple fashions, but a most useful classification is one that separates them along two lines. The first cut divides the conflicts into two groups: one in which the disputants (e.g., two neighbors) will probably interact with each other after the mediation versus those conflicts in which the disputants probably will not interact subsequent to the mediation.

As Table One indicates, the second cut divides the conflicts according to the skill and expertise of the disputants. Some disputants typically have low negotiation/conflict resolution skills, while others are very skilled and sophisticated. As examples, husbands and wives in divorces, two workers on an assembly line, or landlords and tenants in rent disputes probably have less experience in negotiation and fewer conflict‐reduction skills than do union and management representatives, attorneys in civil cases, or salespersons and purchasing agents in interfirm transfers.

This two‐fold division is employed and exhibited because it has implications for the mediation approach that is most likely to be employed: if the disputants will maintain a relationship/interaction after the mediation, it would be reasonable for the mediator to utilize techniques targeted toward maintaining an amicable relationship. These might include calls for apologies, stating the other side's point of view, or suggesting face‐saving proposals. If the disputants will not retain the future relationship, however, then the mediator could focus on obtaining an agreement and could have less concern for the relationship.

Turning to the negotiation skills sector (Table One), if the disputants have inadequate negotiation skills, the mediator would be well advised to use a style that provides guidance for parties. On the other hand, if the disputants do possess adequate skills or are represented by someone who has these — such as in civil cases in which experienced attorneys are at the helm — the mediator could use a style that provides less direction as to how the sides should negotiate.

The second factor in the mediation context/environment is the nation in which the mediation is conducted. The effects of country play themselves out via its culture and institutions. We will consider first the culture and its effects.

As Harry Triandis (1994) and others have noted, culture has two elements: human‐made elements and social behavior. The human‐made elements (i.e., individual differences) consist of the self‐definitions, norms, attitudes, beliefs, roles, social structures, and values that may determine the social behavior. The social behavior is a behavior that occurs somewhat distinctly within a culture group and arises from these beliefs, values, and norms.

Both of these elements can affect the mediation. As for the human‐made elements (i.e., beliefs, values, and norms), they affect the type of approaches that mediators will employ. For example, because mediators in a more collectivistic culture are more likely to believe that conflict is harmful, they will emphasize harmony more frequently than will their counterparts in more individualistic cultures. Specifically, they will more often stress the cost of the dispute to society, have third parties from that society present in the mediation, and ask for their assistance and their advice (Wall, Beriker, and Wu 2010). Likewise, because they value harmony, mediators from collectivistic cultures are more likely to call on disputants to forgive or apologize (Callister and Wall 2004), and these mediators will take steps to save face for the disputants.

Another — quite unfortunate — effect of human‐made elements is to be found in gender biases. In some cultures, women are considered to be inferior and have few rights; therefore, the mediators press them for concessions or do not hear their side of the dispute (Bilefsky 2006; Kouassi 2008; Wall, Beriker, and Wu 2010).

Turning to the effects of social behavior, we find mediators emulate the day‐to‐day behaviors in their culture. For example, if members of a culture use proverbs, metaphors, euphemisms, rituals, and taboos in their communications, then mediators from these societies will rely upon such modi operandi in their mediations (Kouassi 2008). Likewise, if the community (e.g., Israel) has a blunt and assertive communication style, mediators from this society will be blunt and assertive when they mediate (Zarankin and Wall 2007).

Another social behavior, religious practice, can also play a significant role. For example, in Turkey and Malaysia, imams, who often function as mediators, daily engage in religious‐oriented behaviors such as meeting in a mosque, relying on the Koran, and praying with the people (Wall, Beriker, and Wu 2010). In their mediations, they tend to rely on these same behaviors.

Before turning to the effects of institutions, we need to point out that the culture of a society dictates not only how mediators will behave but also who will become a mediator. Frequently, it is difficult to determine why the culture makes its choices. In Turkey, for example, Kurdish tribal leaders seldom mediate; rather, this function is left to the butchers, coffeehouse owners, imams, senior businessmen, and muhtars (minor government officials) (Wall, Beriker, and Wu 2010). In Afghanistan, on the other hand, tribal leaders do mediate (Khapalwak and Rohde 2010). Likewise in Togo, the Yongas' tribal leaders do not mediate, but in the Moba society they do (Kouassi 2008).

Examinations of the effects of mediation institutions have yielded a mélange of findings. For example, the Katarungang Pambarangay mediation system in the Philippines allows the mediators — who generally work as a team — to press the disputants because the disputants must attend the mediation prior to trial without their attorneys present (Tabucanon, Wall, and Yan 2008). In the United States, institutional rules prohibit such an approach. In India, the panchayat (a team of five male elders) institution encourages the mediators to gather information from third parties (Wall, Arunachalam, and Callister 2008). The street committee mediation in China permits mediators to retrain and criticize the disputants (Colatrella 2000). We could continue with about a dozen examples from around the globe that reveal a variety of institutional effects.

To organize these findings, an important step entails a simple division in which we distinguish between countries that either have institutions (e.g., a state‐supported community mediation service and regulations governing the practice of mediation) or do not. (Table Two lists the countries that we determined have formal mediations or do not.) As this table indicates, numerous countries have formal institutions while others rely upon informal/noninstitutional mediations.

As Table Two indicates, formal community mediation is a common practice in most Far Eastern cultures. Buddhist and Confucian communities often rely on mediation to resolve disputes because these societies value social harmony. As a consequence, formal mediation practices exist in China, Hong Kong, Japan, and Korea (Callister and Wall 2004).

We also find that most European countries have developed formal systems of community mediation that have been established in coordination with their legal systems (Martinez‐Pecino, Munduate, and Euwema 2006). These countries have seen the benefit of mediation as an alternative to the courts, which have become bogged down with an increase in litigation. In these countries, disputants often realize the benefit of the speed and frequent agreements that can result from using mediation; therefore, they frequently utilize the process.

In the Middle East, it seems that no Arab nation has an institutional mediation service. Even though many of these communities could support formal mediation, they have not established mediation institutions. Rather, they rely upon respected community members and tribal leaders to resolve disputes less formally (Wall, Beriker, and Wu 2010). Countries in this region without an Arab majority, such as Israel, have developed more formal institutionalized mediation practices (Zarankin and Wall 2007).

Given that some countries have formal mediation institutions and others do not, what is the effect of this difference upon the mediation process? In the current decade, scholars have not addressed this question; therefore, we believe that it would be a fruitful area for future research. It seems reasonable to posit that when formal mediation institutions exist, mediation is more apt to take place, and the mediators in them will be more legalistic in their approach, relying upon rules, laws, precedents, and theories of individual rights. In contrast, when no formal service is present, we hypothesize that mediators would be less assertive and more bound by community norms.

Before discussing recent research on the core mediation process, we, for completeness, will mention those several factors beyond conflict type, culture, and institutions that affect mediation. Some determine whether mediation will take place, while others shape the mediation process.

Often, mediation takes place because disputants have learned about mediation, expect benefits from it, and therefore seek it (Barton 2005; Bates and Holt 2007). Likewise, potential mediators frequently conclude that mediation will benefit them and/or other third parties; consequently, they offer to mediate (Hoge 2007). Some mediators offer their services because they have a personal relationship with the disputants and believe that their assistance will benefit them (Charkoudian and Wilson 2006). On occasion, third parties who are affected by the conflict will force or strongly advise the disputants to utilize mediation and/or pressure potential mediators to provide their services.

As for the factors that shape the nature of the mediation, we noted in an earlier work (Wall, Stark, and Standifer 2001) that time pressure can motivate the mediator to press for agreement, as can the high cost of the conflict and/or the cost of a failed mediation. A mediator's training will also influence the approach that he or she adopts. For example, if a mediator has been trained by an institution that emphasizes problem solving, then the mediator is apt to analyze the dispute. In a similar vein, the mediator's ideology or that of the organization in which she or he operates (e.g., an Equal Employment Opportunity Commission office) will dictate the preferred approach.

It is also worth mentioning that the level of hostility between the disputants can have important impacts on the process. Mediators are more likely to use private caucuses with antagonistic disputants. Similarly, the level of trust between disputants will have an effect, with mediators being more likely to focus on relationship improvement when trust is low (Ross 1996).

In sum, we have noted in this section that recent research has examined extensively the influences of context/environment — containing the conflict type, country, culture, mediation institutions, and other assorted factors — on the mediation process. We now turn to the findings focused on that core process.

As noted in Figure One and the preceding discussion, the context/environment in which the conflict and the mediation occur (including conflict type, culture, and presence and type of mediation institutions) influences the core mediation process. This process, we emphasize, is a dynamic one within which effects will feed back to affect causes (e.g., disputants' anger with a mediator's pressing could feed back to motivate the mediator to terminate the pressing). Therefore, when choosing a model to delineate the process, we should select one that possesses this cybernetic characteristic. One that does so is control theory (e.g., Lord and Hanges 1987; Vancouver 2005), which consists of six phases:

• An individual (e.g., mediator), group, or other system has an objective goal.

• This is compared with the current state or outcome.

• A discrepancy between the goal and the current state is identified.

• The discrepancy leads to discrepancy‐reduction approaches (e.g., choice of mediation techniques).

• These are applied to bring the current state in line with the objective.

• Finally, the current state is compared again with the goal so as to reinitiate or terminate the cycle.

Applying this template to mediation, we note that a mediator, after agreeing to mediate, selects one or more goals (e.g., agreement and/or an improved relationship between the disputants). Having selected his or her goals, the mediator compares them with the current state and finds a discrepancy; for example, the mediator finds nonagreement and a poor relationship.

He or she then decides whether to pursue the goal and how to pursue it. Assuming that the mediator decides to pursue the goal, he or she will continue through the decision process and decide which techniques to use to attain the goal.

Because goals initiate the mediation process, we begin with them. Many current scholars and practitioners consider mediation to be an industrial‐grade Swiss army knife — capable of accomplishing any task — and expect mediators to set their goals accordingly. That is, they expect mediators to have goals of bringing about agreement (Firestone 2009), improving the relationship between the disputants (Craig 2008), attaining social justice (Neves 2009) and social transformation (Li‐On 2009), reducing sexual discrimination (Stallworth, McPherson, and Rute 2001), allowing the disputants to blow off steam (Bleemer 2009), preventing future conflicts (Wort 2009), using efficiency/speed (Zimmer 2010), solving problems (Zwier and Guernsey 2005), forestalling future problems (Kay 2009), reducing future costs (Kloppenberg 2001), reducing violence (Mazadoorian 2009), improving communication between the disputants (Portman 2009), reducing stress (Wiseman 2008), integrating relationships (Stimmel 2002), and overseeing restitution (Umbreit, Coates, and Roberts 2000).

While these are worthwhile goals, it is doubtful that mediators adopt them all when beginning a mediation. Because such a list would be cognitively immobilizing, mediators usually begin simply, with the primary — or highly ranked — goal of agreement (Stevens 1963; Pruitt 1971; Kochan and Jick 1978; Brindley 2006). Quickly, mediators probably then consciously and unconsciously select additional goals based upon their evaluation of the situation, the disputants' behaviors, and inputs from third parties who are affected by the dispute. Influenced by these goals, the mediators choose the techniques or strategies they will employ.

The literature from the past decade — as that from the preceding years — indicates that mediators have approximately one hundred techniques to choose from (e.g., collect information and state an opinion), and researchers have conceptually — but not empirically — categorized them into about two dozen, often overlapping strategies (groups of techniques):

• analytic (e.g., Birke 2000);

• broad/focused (e.g., Currie 2004);

• bottom up (e.g., Mars 2001);

• differentiated (e.g., Regina 2000);

• evaluative (e.g., Lande 2000);

• facilitative (e.g., Gabel 2003);

• insight (e.g., Picard and Melchin 2007);

• mediation‐arbitration (e.g., Ross and Conlon 2000);

• narrative (e.g., Hardy 2008);

• neutral (e.g., Kydd 2003);

• power broker (e.g., Chayes 2007);

• power‐political (e.g., Jones 2000);

• pressing (e.g., Kichaven 2008);

• problem solving (e.g., Harper 2006);

• proper sequenced (e.g., Weiss 2003);

• pragmatic (e.g., Alberstein 2007);

• story telling (e.g., Pinto 2000);

• strategic (e.g., Kressel 2007);

• transformative (e.g., Alberstein 2007);

• transformative‐narrative (e.g., Harper 2006), and

• understanding‐based (e.g., Friedman and Himmelstein 2006).

While this bountiful array of often overlapping and sometimes very similar strategies allows for a thorough description of the available mediation approaches, it can paradoxically retard the advancement of our knowledge. Faced with such a complex set of categories, scholars have not been able to grapple with the two fundamental questions for mediation: (1) What are the major causes/antecedents of mediators' strategies? That is, what causes mediators to use the strategies they do? (2) And what are the major impacts of the mediators' use of particular strategies? (Later in our discussion, we will suggest how those strategies can be folded into fewer categories, so as to better facilitate our understanding of the mediation process.)

We also note that some of these articles on mediation strategies are descriptive, for example, the researchers attempt to determine what mediators do and sort those actions into categories — while others are prescriptive, for example, the authors prescribe what strategies — often new ones — they think mediators should adopt.

Shifting from mediator strategies to mediator–disputant interactions (Figure One), we expected the literature to contain numerous articles — theoretical, laboratory studies, and field research — that focus on the mediator–disputant interactions because this bidirectional interaction lies at the core of the mediation. We did find these works in the past literature. For example, quite early, Henry Landsberger (1955) found that labor mediators adopted a pressing style when the disputants became intransigent. Peter Carnevale and Donald Conlon (1988) reported that the quality of the disputants' offers influences the mediators' recommendations. William Donohue and his colleagues (Donohue, Allen, and Burrell 1985) found that successful mediators utilize assertive techniques when they spot one disputant attacking the other.

But while previous researchers investigated the mediator–disputant interaction, current ones, for the most part, seem to have ignored it. We could identify only three studies in the past decade that report the effect of the disputants' behaviors upon that of the mediator. Zeev Maoz and Lesley Terris (2006) found that mediators use strategies of lower intrusiveness when they perceive that disputants view them as credible. James Wall and Suzanne Chan‐Serafin (2009) found that mediators employ more pressing techniques when the disputants reveal high aspirations. And Kenneth Kressel and Howard Gadlin (2009) reported that the selection of a model for mediation was a function of the mediators' assessment of the disputants' behaviors and attitudes. Three additional authors (Dyck 2000; Holaday 2002; Brazil 2007) have recently advised mediators to consider the influence of the disputants' behavior. But because few studies have explored the disputant–mediator interaction, we currently know little about how the disputants' behaviors affect the mediators' behavior or how mediators' behavior affects the disputants' behavior.

Despite this void, an abundance of authors — in approximately one hundred articles — assume that mediator behaviors affect the behavior of disputants and prescribe the steps that mediators should or should not take to alter the disputants' behavior. For example, mediators are advised to obtain trust (Giovannucci and Largent 2009), be neutral (Greenberg 2003), not be neutral (Svensson 2009), consider cultural differences (Coffey 2006), define the problem well (Benjamin and Irving 2005), evaluate (Kichaven 2008), improvise (Cooley 2007), control emotions (White and White 2001), and disbelieve attorneys (Peters 2007).

These articles, we argue, should be viewed skeptically because they are based on the assumption that the mediator's behavior controls the disputants' behavior. Anyone who has mediated, been a party to a mediation, or observed one knows that is rarely the case. Rather, mediation is an interactive, two‐way process in which the mediators' behaviors affect and are influenced by those of the disputants, and, conversely, the disputants' behaviors affect and are influenced by the mediators' behaviors.

Why is there such a disconnect between so much of the literature and reality? One explanation is that some authors operate with a faulty definition of mediation. To distinguish mediation from arbitration, scholars have defined arbitration as a process in which the third party controls the outcome but not the process. In mediation, the third party is said to control the process but not the outcome. These are catchy definitions, but as Oliver Wendell Holmes once quipped: “A good catch‐word can obscure analysis for fifty years” (Shlales 2007). For mediation, it seems that the catchy definitions have exacted such a price. When one observes an actual mediation, there is an instant realization that this characterization is erroneous; mediators do not always control the process. When a disputant intentionally shows up late, refuses to make an opening statement, demands an offer, throws a catheter across the table, brings up sensitive issues, refuses to discuss issues, intentionally insults the opponent, or walks out of the mediation, the mediator would seem to have very little control of the process.

A second explanation for the erroneous assumption that the mediator's behavior dominates is that investigating ongoing mediations is difficult, time consuming, and expensive — understandably, few disputants and mediators are even willing to allow researchers to “sit in” on their mediations. Therefore, instead of studying the interactions, researchers have relied upon surveys. Subsequently, they have made assumptions about causation and successfully published articles that report mediator behaviors as the independent, causal factors.

Having examined the context/environment of mediation and the mediation process per se, we turn to the most bountiful cluster of reports from the past decade, those examining mediation outcomes. When doing so, we note (in Figure One and Table Three) that these outcomes accrue to the disputants, mediator, and third parties.

A major outcome for the disputants is an agreement that settles or ends the dispute. The current literature indicates that mediation is effective in producing this. In studies prior to 1990, the reported settlement rate was approximately 60 percent (Kressel and Pruitt 1989) and for the 1990–2000 decade, the reported rate on average was 75 percent (Wall, Stark, and Standifer 2001). For our most recent decade, this rate is about the same, roughly an 80 percent agreement rate.

Specifically, in bankruptcy mediation, Julie Kay (2009) reported a settlement rate of 75 percent. In Equal Employment Opportunity Commission (EEOC) mediations, Philip Zimmerman (2001) found 64 percent settlement. In community mediations, Timothy Hedeen (2004) provided 76 percent, and Miriam Landsman and her colleagues (Landsman, Thompson, and Barber 2003) noted 60–80 percent in divorce mediations. (These figures are all for the United States.)

No doubt this high settlement rate is instrumental in providing high levels of disputant satisfaction (Patterson 2006). The literature also indicates that disputants obtain satisfaction from the mediation for several specific reasons: mediation is cheaper (Kloppenberg 2001), swifter (Swendiman 2001), and more enduring (Landsman, Thompson, and Barber 2003) than other forms of conflict resolution. Disputants, it is held, also perceive that mediation is also just: both procedurally (Welsh 2004) and restoratively (Hodak 2004).

In addition to these benefits, the disputants find that the process gives them catharsis (Bleemer 2009) and improves their relationship with the opponent (Crush 2007). Just as importantly, disputants usually view the mediation as fair (Mathews 2004), and it gives the disputants control over the process (Bailey and Robbins 2005) and a heightened motivation to solve their problem (Chen 2006). Because of the aforementioned benefits, the mediated agreements tend to be more thoroughly implemented than are court decisions (Landsman, Thompson, and Barber 2003).

Mediators receive their outcomes principally from the resolution of the dispute. That is, successfully resolving the dispute can help them attain personal satisfaction, prestige, and more work as a mediator (Hoge 2007). Benefits also come directly from the mediation process when the dispute is not settled. At the mundane level, most mediators are paid for their services regardless of the outcome. Mediating also allows mediators to develop (Casella 2000) and hone (Hedeen 2004) their interpersonal skills.

While the literature reports that mediators receive such outcomes from the mediation process, no studies have investigated the mediators' reactions to these outcomes (i.e., the top feedback loop in Figure One). To us, this seems to be a worthwhile arena for study. Some open questions here are: Do third‐party criticisms of the mediated agreement motivate mediators to change their subsequent mediation strategy? Does the lack of agreements lead to strategy modifications and/or does it motivate mediators to forcefully repeat the current ones? And does a reputation for a high percentage of agreements result in mediators' increasing the fee for their services?

Outcomes can accrue to third parties (other than the mediator) from the agreement as well as from the mediation process. The most heraldedagreement outcome for third parties is peace when hostilities draw to a close (Koksal 2006).

As for other outcomes, several studies have indicated that mediation reduces the court docket (Van Epps 2001) and case overloads for government agencies (Swendiman 2001). In divorce mediation, research indicates that the children benefit because mediation reduces the hostility between their parents and leads to agreements that are more favorable for the children (Landsman, Thompson, and Barber 2003). Studies (e.g., Kotlowitz 2008) have also indicated that gang mediation helps to provide safer neighborhoods for the residents.

In sum, a copious number of articles have touted the outcomes of mediation on the disputants, mediator, and third parties. But a close reading of the articles reveals more smoke than fire. As Table Three indicates, more than one half of the articles do not report data; rather, they are based upon the authors' personal experience and conclusions or perusals of other data‐free articles.

Second, almost all of the articles that do contain laboratory or field data have no control or comparison group. A rare exception is a study by Daniel Druckman, James Druckman, and Tatsushi Arai (2004).

The absence of a control group is quite understandable because it is difficult to identify disputants who had the option to use mediation but did not. It is equally difficult to chart the alternative routes these disputants could have traveled. For example, consider civil case mediations of automobile accidents wherein one wishes to study the disputants' satisfaction and the agreement rate. What would be the control group? Such options could include parties who worked out an agreement without filing a suit, who filed a suit but settled before or after a trial was scheduled, who filed a suit and settled after a trial was scheduled, who went to trial and settled during the trial, who completed the trial, and so forth. As the above list indicates, identifying the appropriate control group would be a challenge, which is why so few mediation studies have control groups. Nonetheless, we believe that the effort to work with control or comparison groups would be worth it. Without them, we do not know if mediation is more effective than the alternative.

Consider, for example, agreement rates. Civil case mediations are reported (Kay 2009) to have about a 75 percent agreement rate, which means that 75 percent of the disputants will not go to trial. This sounds pretty good until one considers that across the board, 98 percent of all civil cases do not go to trial, whether parties use mediation or not.

When introducing our mediation paradigm (Figure One), we indicated that mediators have goals and attempt to modify disputants' behaviors so as to accomplish these. The mediators' and disputants' interactions modify the current state and generate outcomes for the disputants, mediator, and third parties.

To fine‐tune our delineation of this paradigm, we emphasize that new goals may emerge for the mediators (e.g., to prevent future conflicts or to punish one of the current disputants) during the mediation. Or their goals may change. In either case, the mediators' behaviors will be altered to pursue such goals.

Furthermore, the mediators' goals might be unconscious. (For example, mediators might unknowingly prefer to pressure disputants; Wall and Chan‐Serafin 2009). And as Kressel and Gadlin (2009) reported, mediators' behaviors are occasionally unconscious and automatic, independent of their selected, conscious goals.

As our model, the recent literature has, for the most part, assumed that mediators are conscious of the goals that they pursue in mediation. It seems that a fruitful area for future research would be to investigate mediators' subconscious goals and the cognitive maps and schema that underpin them, as well as the effects of such goals.

In this article, our primary goal is to present an instructive review of the mediation literature from the past decade, organized in a lucid fashion. To organize this material, we utilize an open cybernetic system model (Figure One) in which mediation is embedded in an environment or context whose factors determine the core mediation process. This process has outcomes that generate feedback to continually sustain or modify its process.

The literature from the past decade, we find, is quite extensive — at least three hundred fifty articles by our count — revealing that mediation has continued its advancement into many arenas. In international, environmental, school, divorce, organizational, consumer, sexual harassment, mental health, debt, insurance, liability, contract, malpractice, victim‐offender, taxation, and intergang disputes, mediation is being practiced, rediscovered, described and renamed, studied, or prescribed.

When we scanned this extensive, diverse, multistrata literature from horizon to horizon, we detected five concentrations of articles. In the contextual/environmental sector (Figure One) we found two groups of articles: (1) those reporting applications of mediation to manifold types of conflicts and (2) indicators that mediation is practiced in many different cultures. Relevant to the core mediation process are two additional groups: (3) extensive lists of mediation goals and (4) those describing the various mediation styles. And we also found (5) extensive reports of mediation outcomes.

The articles in groups one and two are invigorating for they show that mediation is proving useful, expanding, and standing the test of time. On the other hand, we find the replete descriptions of the goals (group three) to be less inspiring.

Here, we must conclude that the authors have been overly expansive. When listing the goals of mediation, scholars and other writers have urged mediators to charge forward with their Swiss‐army‐knife mediation to fix all disputes. They seem to forget that mediation has its limits.

At times, mediation does not work (Smith et al. 2002), which is rather evident in the reported settlement figures. If the agreement rate, for example, is 65 percent, this clearly indicates that 35 percent of the disputes did not settle. Likewise, if 75 percent of the disputants were satisfied with the mediation, then 25 percent were not. Effective or not, mediation takes time and can be expensive; therefore, it can be viewed as an “add on” to current systems for dispute resolution (Goldfein 2006). Mediation can be inaccessible to some parties (Goodman 2006), and even when they access mediation, some groups (e.g., those with less power) are at a disadvantage in the process (Mulcahy 2000; Evans 2001; Kay 2009).

Despite the participants' best efforts, mediation is not always confidential (Sherman 2003; Goodman 2006); therefore, disputants sometimes withhold information, fearing that it subsequently will be used against them. And frequently, mediation does not allow for self‐determination by the disputants because the mediator or active outsiders determine the process and outcomes of the process (Harper 2006). Occasionally, mediation can develop a life and turf of its own within an organization or environment (Bingham 2004). Consequently, mediators can begin to serve their own needs or those of the system rather than disputants' interests (Della Noce 2001).

At times, mediation is viewed as an inferior alternative because the mediator does not have the status of a judge, who would be hearing the case in the courtroom (Bleemer 2009). More disturbing is the perception that mediation is tainted because it reduces transparency and accountability to the general public (Goldfein 2006). For example, it is disturbing to citizens when they learn that the Catholic Church or a major financial institution agreed in mediation to compensate the offended parties but will admit no fault (and presumably will thus make no promise to correct whatever institutional shortcoming may have precipitated the original dispute).

Just as the mediation, the mediators have their warts. They can be distracted by their own concerns and emotions (Minkle, Bashir, and Sutulov 2008), or they can be addicted to their favorite mediation strategy, even when it may be less appropriate than another for the current mediation. They can become overwhelmed by the details or complexity of the mediation, and on occasion, they allow outsiders to influence the process (Menin 2000) and infringe on the disputants' rights (Swendiman 2001).

Perhaps the knee‐jerk response to this literature is that mediation admittedly does have its shortcomings, but high goals stimulate high performance. Perhaps we must caution that the pursuit of unrealistic goals could severely damage the reputation of mediation.

Shifting our focus from goals to mediation strategies, we again find a brimming, but, in our opinion, humdrum list. On the positive side, the descriptions of many strategies provide refined reports of the steps mediators can utilize. And they chart new avenues or approaches (e.g., to equip disputants with conflict resolution skills) that mediators can undertake.

On the downside, these descriptions often seem to be splitting hairs or exploring obscure nuances of formerly recognized strategies. But the primary negative blowback of this excess of description is that it may paradoxically retard our understanding and improvement of the mediation process. Because a major element in the mediation process (Figure One) is the mediator's behavior, every additional strategy that is developed or advocated for investigation spawns a geometric increase in the complexity of the model under study. For example, consider a “direct narrative relationship‐improvement” mediation strategy. Its incorporation into the model necessitates a consideration of the factors that would determine if the strategy is to be used, the factors that affect how it is implemented, disputants' reactions to the strategy, the mediators' response to the disputants' reactions, the outcomes from this interaction, and the feedback from the outcomes. In sum, the mediation process that we are attempting to understand becomes vastly more complicated with the inclusion of each additional strategy.

To simplify the study of mediation strategies, we suggest that scholars should fold these strategies into fewer categories, and there are two ways of doing so. One is the “concurrent usage” approach, which identifies which groups of techniques are used together. A second tack is the “similarity” procedure in which strategies can be merged if they share operational similarities or common goals.

Currently, data on which specific techniques are or should be utilized to achieve each of the various strategies are meager; therefore, a concurrent usage approach is not feasible. But the “similarity” procedure is available because the literature does provide adequate information about the goals of the various strategies. Such a procedure allows us to sort, corral, and consolidate the two dozen strategies into six categories:

• pressing;

• neutral;

• relational;

• analytic;

• clarification; and

• multifunctional.

In a pressing strategy, the mediators use procedures to reduce the disputants' aspirations and limits, to move them off positions, and to nudge them toward agreement. In this strategy, mediators may press the disputants equally or display a bias.

For the neutral strategy, the mediators attempt to establish and maintain an interaction between the disputants. That is, the mediators attend the mediation and set the agenda. Here, they gather information from each side, transfer it — objectively or selectively — to the other, seek clarifications, and note one side's opinion as well as feelings.

While the pressing approach focuses on agreements and the neutral strategy targets interaction maintenance, the relational strategy emphasizes the goals of improved communications, clarification of underlying feelings, and general improvement of the relationship between the disputants.

The analytic strategy focuses on the problem that causes the dispute (e.g., a father is not paying alimony), and its goal is the resolution of this problem.

In the clarification strategy, the mediators' goal is to assist the disputants in clarifying and understanding what they personally want from the mediation.

We have labeled the final strategy as multifunctional because it can be used to pursue any of the above goals. For example, the narrative strategy can be used to obtain agreement, to improve the relationship between the disputants, or to help them understand their goals.

To sort the two dozen strategies, we listed earlier into the six above, we suggest that the pressing, mediation‐arbitration, power‐broker, and power‐political be considered “pressing strategies.” As for the “relational” strategy, we believe that the facilitative, transformative, transformative‐narrative, and proper sequence can be folded into this category. The neutral strategy remains intact, containing only itself. The “analytic” strategy can harbor the analytic, bottom‐up, and evaluative strategies. The “clarification” strategy can host the interest‐based and the understanding‐based strategies. Finally, the narrative, story‐telling, strategic, and broad/focused strategies can be grouped together in the “multifunctional” category.

Our sorting of the two dozen strategies can probably be improved. This admitted, we recommend that researchers — perhaps using a technique employed by Robin Pinkley et al. (1995) — utilize a more analytic method for grouping the strategies.

In addition to condensing the various mediation strategies, scholars should, we argue, terminate the discovery of new strategies — the current ones seem to be working if the measurement of effectiveness is achieving high levels of agreement. Perhaps it is even feasible to condense the strategies to three categories — a settlement‐oriented strategy, a relationship‐oriented one, and a neutral one — then delineate and enhance our understanding of each strategy, and subsequently target the improvement of each.

Moving from strategies to outcomes (the fifth group of articles), we find writers in their expansiveness seem to be rediscovering the windmill. We were pleased to find a stronger coverage of third‐party outcomes than in previous decades, but the reports of outcomes accruing to the disputants and mediator have been redundant with past accounts.

More disappointing than this redundancy is the continued focus on the outcomes of the aggregate overall mediation (e.g., mediation gives a high settlement rate). This global focus is disappointing, if not vexing, for three reasons. Its primary shortcoming, noted previously, is that many reports contain no data. Rather, authors report that the word on the street is that mediation yields reduced violence, high agreement rates, satisfaction, lower recidivism, enhanced prestige, understanding, empowerment, improved relations, high joint payoffs, ethical accords, disputant control, social justice, social transformation, catharsis, balanced power, and prevention of future conflict. Whether these results can be backed up by data remains to be seen.

A second problem is that the reports of mediation outcomes fail to indicate the outcomes of the specific techniques. More specifically, of the hundreds of articles reporting the outcomes of the overall mediation process, only a limited few report the effects from the separate techniques. This raises the question as to whether it matters what the mediators do as long as they are active, keep the disputants at the table, retain a rapport with them, and attempt to smooth the relationship between them.

A third, and somewhat related, unsettling aspect is that authors in their reports of mediation outcomes slide into the marketing mode, selling the merits of a mediation approach they have devised, discovered, or practice. Shifting from the hills to the gullies in the literary landscape, what are some of the studies we expected to find but did not? We would be pleased to find:

• studies of the effects of formal mediation institutions (versus informal mediation) upon the core mediation process;

• observations of mediator–disputant interactions in actual mediations and the establishment of causation direction in the interactions;

• investigations of the antecedents to the mediation strategies;

• the use of comparison/control groups to determine the effects of mediator techniques and strategies;

• investigation of mediators' subconscious goals and the cognitive schema that underpin them;

• reports of the conditions under which mediation is most effective;

• studies of the mediators', disputants', and third parties' reactions to their outcomes;

• more empirical studies (only one‐fourth of the reviewed articles were empirical); and

• less reliance on surveys (which do not indicate causation).

After reviewing the past decade of literature, with its contributions and omissions, what do we conclude and suggest? Our evaluation is mixed: from the application perspective, it is positive, but on the scholarly side, it is rather disappointing. Substantially less progress has been made than in the three previous decades (Wall 1981; Kressel and Pruitt 1989; Wall and Lynn 1993; Wall, Stark, and Standifer 2001) because of replete coverage of the same topics in additional arenas.

In the past decade, scholars have applied mediation to new conflicts and studied it in new locations in which it has found favor from mediators, disputants, policy makers, and the general public. From self‐reports, the verdict has been rendered that mediation is successful; it can take a variety of forms; and with adequate tweaking, it can accomplish manifold goals.

These are valid observations, but they indicate that scholars are redoing the easy work, overlearning old lessons. Now it is time to move forward with structured research programs in which researchers investigate actual mediations, utilizing comparison groups. The comparison group might be a control in which no mediation is utilized or the comparison group could be one in which a different style of mediation is utilized.

When pursuing this course, investigators would be well advised to improve their measures of disputants' behaviors and to condense the variety of categories of mediator strategies that they consider. Currently, the outcome measures of mediation (e.g., agreement, disputant satisfaction, and third‐party benefits) are adequate; therefore, we do not need to discover additional ones. Finally, as noted earlier, researchers should devote more attention to the two‐way dynamics of the mediator–disputant interactions as well as to the feedback effects that the outcomes have upon the core mediation process.

In proffering these suggestions, we do not advocate an unpiloted rush into observational field studies. Rather, we suggest that research be guided by solid theories that can be developed and fine‐tuned by drawing from decision‐making and influence models. This theory building can be complemented with laboratory studies of mediation, which seem to have disappeared in the last decade (Pruitt in press). For such investigations, we suggest using experienced mediators because they differ significantly from college sophomores. While this endeavor would be relatively expensive, it would also be feasible because many mediators travel to a central location (e.g., courthouse, office, and community center) to mediate. This being the case, mediators could be recruited to mediate under controlled conditions. The independent variables could be manipulated via the case descriptions as well as by disputants' rehearsed behaviors. The dependent variables (e.g., mediators' attempts to improve relations) could be recorded with cameras or by observer, and subsequently, the mediators could be interviewed so as to discover their cognitive schema.

A parallel approach would be to convince and train mediators to utilize different mediation styles — such as pressing, problem solving, or relationship building — and then randomly assign them a style for each mediation.

Such field studies, guided and complemented by theory building and laboratory studies, should improve the study of mediation in the next decade. Perhaps they will allow scholars and practitioners to unravel the mysteries of mediation before the biochemists understand aspirin.