The Secrets of Successful (and Unsuccessful) Mediators Continued: Studies Two and Three

This article reports the results of the second and third studies in an ongoing research project designed to determine how mediators succeed in assisting disputing parties to achieve settlements and why they sometimes fail. The research seeks:

• to assist mediators in resolving disputes;

• to improve mediator training by informing trainers of which mediator attributes and skills are most important to successful dispute resolution; and

• to help users of mediation services select mediators by delineating the mediator skills and attributes that mediation users in general regard as important for mediator success.

Participants in mediation can have many different goals. As a result, there are many possible definitions of mediator success. Among the possible goals of mediation are:

• settlement of the dispute that brought the parties to mediation;

• resolution of the underlying conflict that led to the dispute; and

• empowerment of the parties and their mutual recognition of each other.

As a result, a successful mediator may be defined either as

• one who obtains frequent settlements;

• one who typically enables the parties to resolve their underlying conflict(s); or

• one whose mediations typically result in empowerment of the parties and their mutual recognition of each other.

For purposes of this research, we use the first of these definitions — a successful mediator is one who obtains frequent settlements — understanding that regardless of the mediator's talents, disputing parties will not agree to a proposed settlement unless it satisfies the core interests of each and is perceived by each as preferable to its best alternative to settlement (Brazil 2007). While we lack data to measure the number and frequency of settlements achieved by the mediators in this research, we do know that in the commercial, labor, and employment disputes in which these mediators were most frequently involved, parties usually select mediators known for their success in facilitating settlement. Thus we use frequent selection to serve as a mediator as a proxy measure of mediator success.

This is not the first effort to determine why mediators succeed in dispute settlement (see Herrman, Hollett, and Gale 2006 and Wissler 2006 for a wide‐ranging summary of existing mediation research; see also Swaab and Brett 2007). It is, however, among the few empirical studies that draw upon the views of both successful mediators and their clients in order to determine the extent to which their views coincide — or do not — with respect to the reasons for mediator success (see also Hiltrop 1989). It is also the first empirical study of which we are aware to determine, from the perspective of mediation users, which behaviors lead to mediator failure.

In Study One, we asked thirty experienced mediators, nearly all of whom had mediated more than one hundred disputes, how they accounted for their success. “What skills and techniques,” they were asked, “enable you to get settlements? . . . What [do] you view as your essential strengths and techniques?” Seventy‐five percent of the mediators responded that their ability to achieve rapport with disputing parties — a relationship of understanding, empathy, and trust — was central to their success in bringing about settlement (Goldberg 2005). A majority of the mediators attributed their ability to achieve rapport to empathic listening, through which they conveyed the message that they truly cared about the parties' feelings, needs, and concerns. Other mediators attributed their success in achieving rapport to their honesty, ethics, and trustworthiness.¹

The surveyed mediators also reported that once having achieved rapport, their most useful techniques for achieving settlements were to generate novel or creative solutions to the dispute, to display patience and persistence in encouraging settlement, and to use humor to reduce tension.

Study One was limited in that it was based entirely upon the personal observations and reflections of the mediators, with no participation from those who had used their services. Accordingly, in Study Two, we surveyed people who had participated in mediation as representatives of disputing parties (e.g., attorneys) to determine their responses to the question of what led to success in mediation. Then, in Study Three, we asked the same group of disputants' representatives about what constituted unsatisfactory mediator behavior, reasoning that this, too, might illuminate both the key ingredients of mediator success as well as what pitfalls should be avoided.

In order to collect Study Two data, we asked each of the thirty mediators who participated in Study One to provide us with the names of the disputants' advocates (typically attorneys) in six mediations they had conducted — a total of twelve advocates per mediator. Seventeen of the original mediators agreed to do so; thirteen did not.² We replaced those mediators who chose not to participate in the follow‐up study with equally experienced mediators who were willing to provide us with client names.

Of the thirty mediators who participated in the original study, twenty‐eight had mediated at least one hundred disputes; two had mediated between fifty and one hundred disputes. Of the eleven mediators who participated in Study Two but not Study One, ten had mediated at least one hundred disputes and one had mediated between fifty and one hundred disputes. The mediators who participated only in Study Two are thus as successful in the marketplace as were those who participated in both studies.

Most of the twenty‐eight mediators who participated in Study Two deal primarily with commercial, labor, and employment disputes. Two focus on divorce disputes and two on environmental and public policy disputes. Sixteen of the twenty‐eight are male, twelve are female. Four are former judges. The twenty‐eight mediators are based in eleven different states. Many of them have nationwide practices, so their clients are even more widespread.

The twenty‐eight mediators in Study Two provided us with the names of 329 people who had represented disputants in mediations they had conducted.

Each of the 329 disputant representatives (who will be referred to hereafter as “mediation advocates”) was sent a letter stating that we had received his or her name from a named mediator as having been counsel or spokesperson in a recent mediation conducted by that mediator. We asked each recipient of the letter to respond to two questions, with the assurance that we would not share those responses with the mediator in question:

• Thinking back to your most recent mediation with [the named mediator], and any other mediations that you may have had with him/her, what personal qualities, skills, or techniques did [the named mediator] demonstrate that helped move the parties toward settlement?

• How would you account for [the named mediator's] success as a mediator?³

Of the 329 people we surveyed, 216 responded, for a response rate of 66 percent. Seventy percent (152 of 216) of the respondents are lawyers, 22 percent (48 of 216) are union or management representatives in labor dispute mediations, and 8 percent are either representatives of government agencies or public interest organizations in environmental and public policy disputes, or people who represented themselves in the mediation.⁴

We received 47 percent (102 of 216) of the responses in written form and collected 53 percent by telephone. Nearly all the telephone interviews were prearranged. In the course of the telephone interview, which we conducted only if the respondent had not submitted a written response to the questionnaire, the interviewer asked the two questions on the questionnaire, which the respondent had previously received, and transcribed the respondent's remarks directly into the computer. No additional questions were asked, but the respondent was allowed as much time to respond as he/she wished. The only statistically significant difference between the content of the written and telephone responses was that the latter were more likely to refer to the importance of the mediator's evaluation skills.

To code the data we followed several steps. First, Stephen Goldberg read one hundred respondents' answers to both questions, identifying fifteen distinct skills or attributes mentioned by at least one respondent. Goldberg then coded all questionnaires to determine the extent to which those fifteen skills were attributed to each mediator by each advocate who commented on that mediator. In doing so, Goldberg identified five additional skills. He then reread all previously coded responses, noting and coding the additional skills if mentioned.

Next, Margaret Shaw, who knew neither the identity of the respondent nor the mediator, nor how Goldberg had coded the comments about that mediator, independently coded all responses using the same list of twenty skills and attributes developed by Goldberg. Approximately 80 percent of Shaw's coding matched Goldberg's coding; where there were differences, we discussed and resolved them. (To this day, Shaw does not know which responses applied to which mediator nor who provided those responses.)

We next grouped the twenty skills/attributes into three categories in order to place similar skills and attributes together for discussion and analysis. The three broad categories, listed in Table One, are confidence‐building attributes (those mediator attributes that enable a mediator to gain the trust and confidence of the parties), evaluative skills (the mediator's ability to encourage agreement by evaluating a party's likelihood of achieving its goals outside of mediation, typically a prediction of the likely outcome if the matter were decided by a court or an arbitrator), and process skills (those skills by which a mediator seeks to encourage agreement, not including evaluative skills).⁵

We analyzed the Study Two data with a view to answering two questions:

• Which mediator skills or attributes do those who represent disputants in mediations regard as most important for mediator success?

• Are all mediators successful for the same reasons, or are different mediators successful for different reasons?

To accurately measure the advocates' views concerning the skills and attributes of successful mediators, we had to take into account certain data set characteristics. First, some skills or attributes might be mentioned repeatedly by the same respondent. We handled this by counting only a single mention. For example, if a respondent describing Mediator X gave two different examples of Mediator X's being empathic, or one example of empathy and one of friendliness (both of which are in the same code category), Mediator X received only one point for the empathy attribute. Second, some mediators were reported on by more advocates than were other mediators. To prevent the greater number of respondents from increasing a mediator's skill or attribute score, we generated a mean score for each mediator on each skill/attribute. For example, if eight advocates reported on Mediator X, and four of those advocates said that among the reasons for X's success was that he was empathic, X's mean score on empathy would be 50 percent. Similarly, if six advocates reported on Mediator Y, and three of those mentioned Y's empathy as a reason for her success, Y's mean score on empathy would be 50 percent. As a result, the mediators' ratings on each skill or attribute were unaffected by the variation in number of respondents reporting on them.

Next, to determine which skills and attributes were most characteristic of successful mediators in general, we calculated the mean scores of the entire sample of successful mediators on each skill and attribute. This was done as follows. Assume, as noted in the prior paragraph, that both Mediator X and Mediator Y received mean scores for empathy of 50 percent. Assume further that Mediator Z's mean score on empathy was 60 percent, and Mediator A's was 80 percent. If these were the only mediators in the study, the mean score for empathy across all successful mediators would be 60 percent. Thus, 60 percent of the advocates who commented on the average successful mediator thought that being empathic was one of the reasons for mediator success. The results of this calculation are set out in Table One, which shows on average how frequently (the average across all mediators of the average for each mediator) respondents reported a skill or attribute as being characteristic of a successful mediator. These data were used to answer our first research question: which skills or attributes do mediation representatives view as most important for mediator success?

As shown in Table One, the most frequently cited behaviors correlated to mediator success involved the mediator's ability to gain the confidence of the parties, albeit by different means. Tops on the list — referred to by an average of 60 percent of the mediation advocates commenting on the average successful mediator — was that the mediator was friendly, empathic, likable, etc. Examples of the respondents' comments include:

The next most frequently cited reason for mediator success — referred to by an average of 53 percent of the mediation advocates — was that the mediator had high integrity, as demonstrated by his/her honesty, neutrality, trustworthiness, protection of confidences, etc. Examples of these comments include:

Rounding out the top three most frequently cited reasons for mediator success — referred to by an average of 47 percent of the mediation advocates — was that the mediator was smart, well prepared, and/or knew the relevant contract or law. Examples of the responses include:

The results in Table One support the conclusion that an essential attribute of the successful mediator is his/her ability to gain the confidence of the parties. This result, based on advocates' responses, is consistent with the views of the mediators themselves, as reported in Study One (Goldberg 2005). But in Study One a majority of the mediators attributed their ability to gain the confidence of the parties to their ability to listen empathically, which led the parties to like and trust them, and only a few mediators attributed their ability to obtain the confidence of the parties to their honesty and/or integrity. In the present study, advocates regarded both attributes — empathy and integrity — as central to mediator success.

Another difference between the responses was the greater frequency with which advocates in Study Two mentioned the mediator's intelligence and preparedness as factors in his/her success. We think a possible explanation for these differences is that successful mediators, the respondents in Study One, may assume that integrity, intelligence, and preparation are so obviously and self‐evidently crucial for achieving success as a mediator that such traits are not worth mentioning.

Table One also indicates that the mediators' confidence‐building attributes were cited by respondents more frequently than were the various skills used by mediators to bring about agreement. The most frequently mentioned mediator skills/attributes were patience and persistence (referred to by an average of 35 percent of the mediation advocates), providing useful evaluations or reality testing regarding the likely outcome of the dispute in court or arbitration (33 percent), and asking good questions and listening carefully to responses (28 percent).⁶

Some comments relating to the mediator's patience and persistence include:

Comments involving the mediators' provision of useful evaluations or reality testing regarding likely outcomes in court or arbitration include:

Comments involving the importance of asking good questions and listening carefully to responses include:

The respondents also valued the following skills/attributes:

• Diplomacy and tact (21 percent):

• Proposing solutions/being creative (19 percent):

• Keeping the parties focused (15 percent):

• Being candid/firm as necessary (15 percent):

• Understanding people and/or relational dynamics (13 percent):

• Being calm and/or deliberate (12 percent):

A comparison of the advocates' views as shown in Table One with the views of the mediators who participated in Study One reveals both similarities and differences. The two skills that both the mediators and the advocates agreed were important were being patient and persistent, and proposing solutions and being creative. They differed notably in the importance they assigned to mediator evaluation skills, a factor regarded as important by 33 percent of the advocates but by fewer than 10 percent of the mediators. Advocates then, appear to regard evaluation skills as more relevant to mediation success than do the mediators themselves.

In order to generate an overall score for each mediator, we aggregated across process skills to create a single category labeled “process skills.” To do this, we summed a mediator's mean scores on each of the sixteen process skills and divided by sixteen. This gave us a score on the process skills category that was comparable to the mediator's scores on evaluation skills and each of the three confidence‐building attributes (friendly/empathic, high integrity/honest, and smart/well‐prepared).

We next standardized all mediators' scores on the five skills and attributes, summed the standard scores and restandardized. Standardization generates a mean of zero and a standard deviation of one across mediators on each of the five skills and attributes and on the overall performance score.

Standardizing the mediators' scores on each skill and attribute enabled us to

• compare a single mediator's scores on each of the five skills and attributes in order to see where he/she was strongest;

• compare different mediators on the same skill or attribute;

• create an overall score for each mediator that was equally weighted for each of the mediator skills and attributes.

Table Two lists the skills/attributes profile and relative standing of twenty‐six of the twenty‐eight mediators in Study Two.⁷ Relative standings are based upon each mediator's standardized score on each of the skills and attributes: friendliness/empathy, honesty/integrity, intelligence/prepared‐ness, process skills, and evaluative skills.

The symbols in Table Two indicate, for each skill or attribute, whether the mediator was at or above the mean for all twenty‐six mediators on the skill or attribute (+), at least one standard deviation above the mean (++), or below the mean (0). For example, Mediator G was at or above the mean on friendly/empathic, smart, and evaluation skills; at least one standard deviation above the mean on honesty/integrity; and below the mean on process skills.

We must make two important points regarding the results displayed in Table Two. First, we considered the possibility that a high overall score for one mediator, when compared to another mediator, might mean that the respondents who commented on the higher‐rated mediator were simply more verbose than those who commented on the latter. If there had been only one or two respondents per mediator, this would have been a legitimate concern.

As noted, however (see Note 7), the average number of respondents per mediator was eight, and no mediator whose assessment was includedin Table Two was commented on by fewer than four respondents.Furthermore, the responses for each item in Table Two were averaged across respondents. Thus, a mediator could not receive a high comparative score on a Table Two skill or attribute unless a high proportion of his/her respondents credited the mediator with possessing that skill or attribute. Accordingly, we are confident that the differences in the mediators' overall scores do not reflect differences in the verbosity of the respondents, but represent genuine differences in the respondents' perception of each mediator's skills and attributes. We are similarly confident, because of the substantial number of respondents per mediator, that a high score on a particular skill or attribute does not merely reflect the fact that individual respondents might be particularly sensitive to certain skills or attributes, such as intelligence or “niceness.”⁸

Second, the mediators with the highest overall Table Two scores are not necessarily “better” or “more successful” than the other mediators — they are all successful in marketplace terms and, we believe, all successful in dispute resolution terms. Table Two does, however, explain why the mediators in that table are successful. We know, for example, that, in addition to Mediator B's other qualities, a high proportion of the advocates who commented on him/her view Mediator B as friendly/empathic and providing useful evaluations. Similarly, we know that Mediator L is viewed by a high proportion of his/her respondents as friendly/empathic and possessing process skills. Indeed, as one moves down Table Two, one can discern, from the respondents' perspectives, the reasons for the success of nearly all the mediators in Table Two. To be sure, some of the mediators lower down in Table Two were only at or above the mean on one or two skills and attributes, but in this select group of highly successful mediators, that may be enough to be successful — or the lower‐ranked mediators may possess skills or attributes of which their respondents were unaware or that they failed to note in our study, but which account for their success.

What we find most striking in the results displayed in Table Two is that no single profile characterizes each and every one of the successful mediators, much less of those mediators whose overall scores were the highest. Nor are the highest‐ranked mediators outstanding with respect to all skills and attributes. Mediators A and D are the only mediators who were outstanding — more than one standard deviation above the mean — in more than two categories. Only seven (including A and D) of the twenty‐six mediators in Table Two were more than one standard deviation above the mean on more than one skill or attribute.

We found no significant correlation between a mediator's gender and that mediator's overall score or that mediator's scores on any of the five skills or attributes. Female mediators were not cited significantly more or less often for being friendly and empathic than were male mediators, nor were female mediators cited significantly more or less often for their process or evaluative skills than were their male counterparts.

Nor do our results reveal any significant difference between the overall evaluations or individual skills/attributes scores of the four mediators who were former judges compared to those mediators without judicial experience. The former judges were neither significantly more often cited for their evaluation skills nor significantly less often cited for their process skills than were other mediators. To be sure, neither the four former judges who participated in this study nor the other mediators who did so are representative of all practicing mediators. Each of the mediators in Study Two is highly successful, and it seems likely that the process skills of the four former judges play some role in their success. In brief, whatever merit there may be to the view that former judges are more highly valued as mediators for their case evaluation skills than for their process skills, we found no support for that view among this small sample of highly successful mediators.

The only significant correlation between a mediator's score on one skill or attribute and that mediator's score on another skill or attribute is found in the relationship between the mediator being viewed as smart, well prepared, knowledgeable about the relevant contract or law, and the mediator being viewed as providing useful outcome evaluations. Not surprisingly, those mediators who received high scores on smart/well prepared/knowing relevant contract or law were significantly more likely to receive high scores for providing useful outcome evaluations — typically a function of knowing the relevant contract or law.

We also note that eleven of the thirteen mediators (Mediators A through M) whose overall scores are above the standardized overall mean score (all except for Mediators J and K) are more than one standard deviation above the mean on at least one of the confidence‐building attributes. This, we think, corroborates the Study One finding about the importance of confidence‐building attributes for mediator success.

Study Two indicates that, from the perspective of mediation advocates, the most important attributes of successful mediators are those that build the disputants' confidence in the mediator. These are:

• friendliness and empathy;

• honesty and integrity; and

• being smart, well prepared, and/or knowing the relevant contract or law.

The most important of the successful mediators' process skills from the advocates' perspective, are:

• proceeding with patience and persistence;

• providing useful evaluations and/or “reality checks”; and

• asking good questions and listening carefully to the responses.

We also found in Study Two that the advocates view different mediators as achieving success as a result of different combinations of skills and attributes. Some of the mediators with the highest overall scores were rated as outstanding — more than one standard deviation above the mean — in the categories of being friendly/empathic and possessing excellent process skills or evaluative skills; others were rated as outstanding for possessing high integrity and excellent process or evaluative skills; while still others were rated as outstanding in the categories of being smart, well prepared, knowing the relevant contract or law, and possessing excellent evaluative skills. The sole characteristic shared by nearly all the thirteen mediators in the top half on the overall advocate scores was that eleven of the thirteen were a standard deviation above the mean on at least one of the confidence‐building attributes.

The fact that all the mediators in Studies One and Two are successful can be seen in some respects as a weakness of those studies. In Study One, we could not compare the views of successful mediators concerning their skills and attributes with the views of less‐successful mediators. Nor, in Study Two, could we compare the respondents' views of the skills and attributes of successful mediators with their views of the skills and attributes of unsuccessful mediators — no unsuccessful mediators were included in Study Two. Study Three is, therefore, an attempt to compensate, at least in part, for this weakness, by exploring the views of mediation advocates concerning the ways in which some mediators (not those participating in Study Two) failed to satisfy their expectations.

Each of the 216 mediation advocates who responded to the Study Two questionnaire was sent a second letter, which contained two additional questions:

• Have you ever participated in a mediation in which the mediator engaged in conduct that you thought was counter‐productive, that reduced the likelihood of settlement? If so, what was that conduct?

• Have you ever participated in a mediation during or after which you decided that the mediator was so unsatisfactory that you would never again use that mediator? If so, why? What personal qualities or behaviors of the mediator led you to that conclusion?⁹

This letter did not refer to the mediator who had originally provided us with the respondent's name. Additionally, the respondent was requested not to report the name(s) of the mediator(s) whose behavior was described. This request was almost universally honored. In the few instances in which the respondent did include the name of the mediator to whom he/she was referring, the mediator was not one included in Study One or in Study Two.

Of the 216 persons to whom the Study Three questions were sent, ninety‐six responded for a response rate of 44 percent. Seventy percent of the respondents were lawyers, 23 percent were union or management representatives in labor dispute mediations, and 7 percent were either representatives of government agencies or of public interest organizations in environmental and public policy disputes, or people who represented themselves in mediation. The distribution of respondents among these three groups — lawyers, union/management representatives, and others — was almost identical to the Study Two distribution.¹⁰

We received 73 percent of the responses in written form; we obtained the remainder in telephone interviews in which, as in Study Two, the interview took place on a prearranged date and time, the interviewer repeated the questions on the questionnaire previously sent to the respondent, asked no questions other than those on the questionnaire, allowed the respondent as much time to answer as he/she wished, and transcribed the respondent's remarks directly onto the computer. The 73 percent proportion of written responses was greater than in Study Two, in which 47 percent of the responses were written. We suspect that the difference is attributable to the fact that the people to whom we sent the Study Three questionnaire had already participated in Study Two; hence, fewer of them needed the prodding of a follow‐up telephone call to respond.

The only statistically significant difference between the content of the written and telephone responses was — as in Study Two — that the latter were more likely to refer to the mediator's evaluation skills. Neither here nor in Study Two can we account for this difference. The questions asked in the oral interview were exactly the same as those in the written questionnaire; nothing was asked during the oral interview that was not asked on the written questionnaire. Why there were more references to evaluation skills in the responses to the oral interview than in the responses to the written questionnaire in both Studies Two and Three remains a mystery — happily not one that would appear to undercut the results of either study.

In coding the Study Three criticisms of mediators, we used antonyms of the positive skills/attributes codes developed in Study Two. For example, the Study Two code “friendly/empathic” becomes “self‐absorbed/not empathic” in Study Three. Similarly, the Study Two skills/attributes groupings — confidence‐building attributes, process skills, and evaluation skills — become lack of confidence‐building attributes, lack of process skills, and lack of evaluation skills.

As in Study Two, all coding was done initially by Stephen Goldberg, then independently by Margaret Shaw. Approximately 80 percent of Shaw's coding was the same as Goldberg's; disagreements were resolved by discussion. Both Goldberg and Shaw knew the identity of the respondents; neither, except in one or two instances, knew the identity of the mediator whose conduct was described. As noted previously, the few mediators who were mentioned were not among the mediators involved in the earlier studies.

Approximately one quarter (23 percent) of the respondents reported that they had never observed a mediator engage in counterproductive conduct and had never used a mediator whose conduct was so unsatisfactory that the respondent would never again use that mediator. As shown in Table Three, the most common criticism of the unacceptable mediator, reported by 48 percent of the respondents, was that the mediator lacked integrity.¹¹

Some of the reported behavior struck us as nearly incredible:

The absence of other confidence‐building attributes was also the basis of considerable criticism. Twenty percent of the advocates criticized mediators who lacked empathy, and appeared more interested in themselves than in the parties. Respondent's comments included:

Sixteen percent of the respondents commented that the mediator did not understand the issues or the law, and/or was not well‐prepared. Among their comments:

The process skills failure that was far and away the basis of the most criticism, referred to by 24 percent of the advocates, was that the mediator was not forceful in seeking a settlement, but just went through the motions of mediation, doing little more than carrying messages back and forth between the disputing parties:

The most striking omissions from the Table Three critical comments, when compared with the Table One list of positive mediator attributes and skills, are the comparatively rare criticisms of mediators for not being patient/persistent (reported by 11 percent of the Study Three advocates, compared to 35 percent of the Study Two advocates who regarded patience and persistence as characteristic of successful mediators) and the similarly rare criticisms of mediators for poor evaluative skills (reported by 7 percent of the Study Three advocates compared to 33 percent of the Study Two advocates who regarded good evaluative skills as an important element of mediator success).

Similar differences are found with respect to the criticism of mediators for not asking good questions/listening carefully (mentioned by no Study Three respondents but said to be important to mediator success by 28 percent of the Study Two respondents) and not being tactful/diplomatic (also mentioned by no Study Three respondents but characterized as important by 21 percent of the Study Two respondents).¹²

We suspect that the reason for the comparatively low frequency of these criticisms by the Study Three advocates is because the absence of these skills and attributes pales into insignificance when compared to the central Study Three criticisms:

• that the mediator lacked integrity, cared more about himself/herself than resolving the dispute, or was unprepared/uninformed about the relevant issues and/or law; and

• that the mediator did not demonstrate any process or evaluative skills, but was merely a messenger, transmitting messages from one party to the other.

Faced with these behaviors, it is hardly surprising that the respondents went no further in their criticisms, and their failure to do so is not necessarily inconsistent with the views of the Study Two advocates concerning the importance of skills such as patience/persistence, tact/diplomacy, asking good questions/listening carefully, and being capable of providing useful outcome evaluations.

Although Study Two indicates that a wide range of attributes and skills can contribute to a mediator's success, Study Three indicates that lack of success is primarily attributable to a few fundamental flaws. The most important of these flaws is a mediator's lack of integrity, demonstrated by such conduct as unauthorized disclosure of confidential information, providing the parties with inconsistent evaluations of the likely outcome of the dispute, and showing a lack of neutrality. Other mediator conduct widely viewed as a recipe for failure includes not demonstrating genuine interest in resolving the dispute; not understanding the issues involved in the dispute and/or being unprepared; and doing little to assist the parties to achieve resolution, that is, doing nothing but carry messages back and forth between the parties.

The central conclusion to be drawn from these three studies is that a — if not the— core element in mediator success is the mediator's ability to establish a relationship of trust and confidence with the disputing parties. Most of the Study One mediators thought that achieving such a relationship was a result of their convincing both parties that they truly cared about the parties' needs and concerns; a few attributed their success to their honesty, strong ethics, and trustworthiness. The advocates in Study Two, however, assigned essentially equal importance to these different attributes as well as to the mediator's knowledge and preparedness, suggesting that mediator success in gaining the trust and confidence of the parties is equally likely to be associated with any of these attributes.

Both the mediators in Study One and the advocates in Study Two regarded persistence and creativity as important for mediator success. Neither of those skills, however, was as widely regarded as important by either the mediators or the advocates as were those attributes that we have characterized as confidence building.

Study Two also suggests that different mediators can be highly successful on the basis of different types of skill sets — process skills seem to be key for some mediators, evaluation skills seem to be key for others — and nearly all highly successful mediators are widely viewed as possessing at least one of the confidence‐building attributes.

Study Three approaches the reasons for mediator success from a different perspective — asking why some mediators are not successful. The Study Three results reinforce the conclusions of Studies One and Two regarding the importance of obtaining the confidence of the parties. According to the advocates who responded to Study Three, the most common cause of mediator ineffectiveness was that the mediator lacked integrity — he/she disclosed confidences, gave inconsistent evaluations, was biased, etc.

Few of the Study Three respondents viewed a lack of mediator skill as a central element in the mediator's lack of success, with one prominent exception. Not surprisingly, the Study Three respondents reported they would be unwilling to use a mediator again if that mediator contributed essentially nothing to the search for a resolution to the parties' dispute other than to relay messages from one party to the other.

The common theme running through Studies One, Two, and Three, then, is that gaining the trust and confidence of the parties is the most important element in mediator success. The mediator's skills are also important, but these were less often cited as reasons for mediator success than were the mediator's confidence‐building attributes. Finally, and of considerable importance, there is no single model of the successful mediator. Different mediators succeeded on the basis of different combinations of attributes and skills.

Perhaps the most important finding of this research for the practicing or aspiring mediator is that the key to mediation success is quite straightforward:

• obtain the trust and confidence of the disputing parties by being friendly and empathic, by demonstrating high integrity, or by being intelligent, well prepared, and/or knowledgeable in the relevant law or contract; and

• be capable of taking advantage of the trust and confidence of the parties to assist them in resolving their dispute by exercising one or more of the skills set out in Table One.

Some aspects of achieving success as a mediator can be achieved by training, but others cannot:

• The mediation trainer cannot train aspiring mediators to be smart or to know the relevant law or contract, but he or she can emphasize the importance of being well prepared for mediation. Similarly, the trainer can emphasize the importance of integrity, for example, by using simulations to put trainees in situations in which they are tempted to act inappropriately by breaching confidence in the hope that doing so will aid in obtaining a settlement.¹³

• The instructor cannot teach empathy — the mediator's genuine concern for the needs of each party — but he or she can teach ways of showing genuine concern through demonstrations and interactive exercises (Goldberg 2005).

• Many of the process skills set out in Table One can and are being taught and practiced in mediation training. Although some aspiring mediators will demonstrate greater aptitude for some of these skills than for others, it is worth remembering that the most widespread criticism made by the Study Three advocates was not of mediators who lacked a particular skill, but of the mediators who were perceived as doing nothing to assist the parties other than relaying messages.

Finally, the findings of these studies could be useful to advocates, such as attorneys and labor negotiators, who engage in mediator selection. It is commonplace for advocates in search of a mediator to inquire about a particular mediator from others who have used that mediator's services. Most often, the inquiry consists of asking, “How good a job did X do for you in the ABC mediation?” or words to that effect. Based on this research, however, we advise advocates to ask more pointed questions relating to the mediator's empathy, integrity, knowledge of the relevant contract or law, persistence, etc., focusing on those skills or attributes that the advocate believes would be most useful in resolving the particular dispute for which a mediator is being sought (See Sander and Goldberg Forthcoming).

In sum, a better understanding of the attributes and skills of successful (and unsuccessful) mediators can be useful in improving the practice, teaching, and selection of mediators.

The findings of Studies Two and Three suggest at least two future studies. In Study Two, we found that former judges were not significantly more often cited for their evaluation skills, nor significantly less often for their process skills, than were mediators without prior judicial experience. Similarly, we found no significant difference between former judges and other mediators in the frequency of respondent references to the mediator's empathy and friendliness as compared to the mediator's intelligence and knowledge of the law. There were, however, only four former judges in the Study Two sample, far too few to be entirely confident that a larger sample might not show that the success of former judges in the mediator role might be more dependent on their knowledge of the law than on their empathy and friendliness, and more dependent on their evaluation skills than their process skills. Thus, a future study that explores these questions in a larger sample of former judges might be valuable.

Another potentially valuable future study would test whether the same skills and attributes that appear to account for the success (or lack of success) of mediators in the United States are equally important in other countries in which mediation is frequently used as a dispute resolution process. Culture is a powerful factor in how people deal with conflict (Brett 2007). Is it an equally powerful factor in the skills and attributes needed by a successful mediator to resolve conflict? A future study that applied the techniques of Studies Two and Three to examine the reasons for mediator success in another culture might shed light on that question and be as useful to mediators and mediation advocates in other countries as we hope Studies Two and Three will be in this country.