Mediators’ Work: What Do Empirical Studies of Persuasion Tell Us?

Vol. 18 No. 1


Reader Advisory: This article (the first in a two-part series) is about the use of persuasion by mediators to help disputants achieve resolutions they seek but that might otherwise elude them. If this resonates with your concept of the mediator’s proper function, what follows should be useful.  But if you are one for whom the words “mediator persuasion” trigger discomfort, these articles may be especially important to read.

Suppose that a law-trained mediator is mediating an employment discrimination dispute. The 59-year-old plaintiff alleges that she was first passed over for a promotion in favor of a younger male, and then fired six months later from her office manager position because of age and gender discrimination. The company, a manufacturer of seamless gutters, denies the charges, claiming in its answer that the plaintiff was laid off because of economic exigencies caused by the poor housing market. Although her prior work evaluations were all good and there have been several provable incidents of sexist workplace remarks by supervisors, the plaintiff’s case is anything but a slam-dunk. For one thing, she made a scene in front of her co-workers when she was passed over for promotion, referring to her supervisors in ways that even she now concedes were “unprofessional.”

Both sides are represented by counsel at the mediation, which has been in progress for more than six hours. After an initial settlement demand of $200,000, a lengthy and heated factual exchange, and a wide variety of non-evaluative mediator interventions, the parties seem tired, angry and are still $45,000 apart. The parties’ attorneys have vastly different assessments of the strength of the plaintiff’s case, in large part because they disagree about whether the episode in which the plaintiff publicly “disrespected” her supervisors will be admissible as evidence. The plaintiff’s attorney argues that this episode is irrelevant and inadmissible because the defendant stated, in its answer to the complaint, that “purely economic considerations” prompted the company’s actions.  Calling this “hyper-technical,” the company responds that a host of job performance factors–including the plaintiff’s mercurial temperament and expressions of job dissatisfaction–were part of the “economic considerations” that led to its decision to let her (and not someone else) go.

Persuasion: Questions vs. Statements
The mediator is almost sure that the plaintiff’s position on this question is incorrect because a recent Court of Appeals decision rejected a similar argument  (although there is a slight chance the case could be distinguished). She also thinks that persuading the plaintiff on this point–or at least creating some doubt in her mind–is crucial in bridging the bargaining gap. To succeed in doing so, she is considering two approaches:

        (A) Telling the plaintiff and her lawyer that their argument is very likely a losing one. (“From where I sit, this incident will almost surely come into evidence. And, if it does, you’re looking at a greatly weakened jury appeal in these tough job times.”)

        (B) Asking the plaintiff and her lawyer questions, including rhetorical questions if necessary, that suggest the legal risks of their position. (“Aren’t you concerned about the recent Court of Appeals decision?”…Hmmm…. But if the incident does come into evidence, what will that do to your chances?”)  

If you prefer alternative (B), you have lots of company. The technique of persuading by asking questions is deeply rooted in Western rhetorical tradition. Many mediation theorists echo this view. For example, Professor Lela Love, in discussing the mediator’s duty to assure that participants are making fully informed decisions, has written that “asking questions [regarding possible legal outcomes] comports with the mediator’s role, but giving or suggesting answers does not.”[i]

Why do many mediators prefer asking to telling when trying to persuade? For some, questions feel less argumentative or pressuring than statements, and more consistent with norms of mediator neutrality and party self-determination. Some fear that making statements about probable legal outcomes is tantamount to giving legal advice, which they view as inconsistent with the mediator’s role. Asking questions may be attractive to some as a way to sow doubt without having to make definitive predictions in an unpredictable legal world. Still others assume that questions, rather than statements, make the subject more likely to engage with the message and/or less likely to resist it.

Practicing mediators are more or less comfortable with the use of persuasion, depending on their training, philosophy and the settings in which they work. But most practitioners know little about what might actually be effective in mediation persuasion–and why. Fortunately for the field, there is a wealth of data on which we may be able to draw. In the past 50 years or so, social scientists have made substantial inroads in demonstrating empirically what kinds of persuasive interventions work, in contexts ranging from consumer advertising and public health campaigns to jury advocacy and politics.[ii] We have analyzed some of that research in order to find out what it might tell us about a number of common mediator interventions.

So, does this research provide any guidance to our discrimination claim mediator–and mediators in general–as to which mode, telling or asking, might be more effective in changing attitudes?  If mediators seek to influence disputants’ thinking in a non-pressuring way, empirical research suggests that delivering a straightforward message may be more effective, and that suggesting or embedding a message in questions may be counterproductive. Here’s why:

Questions may interfere with message processing. Although some research suggests that questions tend to produce more elaboration (thinking) about a message (and, thus, more persuasion if the message is strong) than statements do, when people are “highly involved” –i.e., personally affected by and already deep in thought about the topic, as mediation participants tend to bequestions may serve as a distraction from their engagement with the message and actually reduce thinking about it. (This was the finding in a well-known study of college students considering whether a new comprehensive pre-graduation exam should be required of them at their campus.)[iii]     

Questions can focus attention on the persuader in ways that undermine persuasion. For people who have any sophistication about persuasion tactics, the use of questions instead of more straightforward statements comes as a deviation from what they expect. This, in turn, can lead them to focus more on the message source and her tactics than on the message itself. And when their assessment of the persuader is unfavorable, those subjected to suggestive questions are less likely to be persuaded–and more likely to feel pressured–than individuals who received the same message in statement form. Indeed, in a pair of studies replicating the college exam study, questioner-persuaders were perceived as less expert, and no more polite, than their statement-making counterparts.[iv]

Statements: One or Two-Sided? How Explicit?  
Even if direct statements (as in our example, about the Court of Appeals ruling and its predicted impact on the plaintiff’s case) are the preferred mode of delivering “bad news,” there are a range of choices about how to frame such statements. For example, the mediator might provide a one-sided view of why the Court of Appeals ruling will apply.  Or she could outline the arguments on both sides.  If she presents both sides, she could add the reasons that make the other side less persuasive. She might provide more or less explicit detail in support of her prediction. She could also express her conclusion clearly and specifically or leave it implicit or general. Does social science research shed any light on these variables?

    ·        One-Sided Versus Two-Sided Messages. When it comes to direct statements, a 1999 meta-analysis of many individual studies concludes that there is, overall, “no dependable difference in persuasive effectiveness between one-sided and two-sided messages.” However, refutational two-sided messages–messages that explain why one side of an argument is more likely to prevail than another–enjoy “a dependable persuasive advantage over one-sided messages.” By contrast, non-refutational two-sided messages are “significantly less persuasive than their one-sided counterparts.”[v] 

    ·        Explicit Versus Implicit Argumentation. Argumentative explicitness is also persuasive. Specifically, arguments that expressly reveal their sources and/or provide more explicit argumentative support “are significantly more credible and significantly more persuasive than their less explicit counterparts,” according to the research.[vi]

    ·        Explicit Versus Implicit (or Missing) Conclusions. Finally, empirical research demonstrates that messages with explicit overall conclusions are significantly more persuasive than messages without conclusions, and that messages with more specific conclusions are significantly more persuasive than messages with generalized conclusions. These findings seem to hold true irrespective of subjects’ pre-existing (favorable or unfavorable) viewpoints.  There is thus no reason to suspect that these effects are different when message recipients are “dug in” against a persuasive message.[vii] 

Applied to our example, then, the most persuasive way that the mediator might present her viewpoint would be to provide a detailed, fully-supported explanation of her views and conclusions, for example: “One could arguably conclude that the Court of Appeals decision does not apply here.  But unfortunately I think that the contrary case is far more compelling.  The Court stated specifically that ‘[x]’…. Two other District Court decisions have, citing the Appeals Court decision, admitted evidence like this in almost identical circumstances (describing cases )….Now let me talk specifically about how a jury might react to this evidence and the impact I fear it would have on your ability to convince them that you were discriminated against….”    

Limitations and Caveats
In reporting these findings in necessarily condensed form, several caveats are in order. First, the research on rhetorical questions is not as robust as the research on message sidedness or explicitness. For example, researchers have not agreed on any common definition for rhetorical questions, and different studies have utilized different forms, placements and numbers of rhetorical questions in an argument, all of which may have affected their results.

Second, in making statements such as “[x-type] interventions have been shown to be more effective in persuasion than [y-type] interventions,” we obviously do not mean to suggest that any particular persuasive intervention will or will not work in a specific situation.  Social scientists deal in generalities, not particularities.  Moreover, theorists agree that persuasion is usually an incremental process, in which people’s minds are changed gradually and by degrees, by means of multiple interventions over time.[viii]

Third and most fundamentally, the studies we canvas are of persuasive interventions in fields as diverse as advertising, disease prevention, race relations and politics; virtually none are of persuasion in mediation.  Making reliable generalizations from any body of empirical work, no matter how broad or sound, is a dicey proposition.[ix] While we are obviously interested in speculating how this research may apply to the world of mediation, we also want very much to avoid overly broad claims. Clearly, research focused on persuasion effectiveness in the specific context of mediation would help better inform our field. 

Lessons and Challenges for Mediators?
Notwithstanding these limitations, we believe that these findings may offer illuminating insights that could inform important debates in mediation. But to let the light in, the field must first come to terms with its squeamishness about the very subject of this research. In many mediation quarters, the word “persuasion” (as opposed to the friendlier “problem-solving”) has a pejorative connotation, with images of slick or heavy-handed neutrals seeking to get a deal at almost any cost.

But most mediators engage in a broad range of persuasive activity, ranging from encouraging participants to participate fully and constructively in the process, to underscoring opportunities for newfound empathy, to serving as “agents of reality”, to more overt forms of evaluation. Writing more than 15 years ago, Deborah Kolb and Kenneth Kressel observed that because mediators are frequently uncomfortable about the persuasive powers they exert, they are prone to engage in a “kind of denial about what they do. [This] denial stands in the way of learning and keeps the field from better understanding the uses and limits of pressure.”[x] Given the pervasiveness of persuasive behavior by mediators of all orientations, the time seems ripe to confront this barrier to our growth as a profession.

So what might be the lessons of this research for mediation practice and norms? First, there is the much-debated question regarding the relative importance of substantive (as opposed to process) expertise as a mediator qualification.[xi] If highly detailed and explicit arguments and conclusions are more effective in persuasion than less explicit or detailed ones, this suggests the importance of legal or industry knowledge as a key criterion for the mediator whose success may depend on changing others’ minds.

Second, these lines of research may shed helpful light on longstanding theoretical debates about facilitative versus evaluative mediation. In mediation, many who seek to persuade eschew explicit or direct efforts at persuasion, often in the name of party self-determination. Even when feedback has been expressly requested by the parties, mediators may “hedge their bets” by giving half-baked evaluations or making statements (if at all) in an indirect, roundabout fashion, even regarding questions that have clear answersout of fear of being seen as favoring a side, appearing coercive, or otherwise acting “inappropriately.”  At a minimum, this body of research may call into question some of the receivedbut untestedwisdom in the field about the kinds of mediator approaches that enhance autonomous decision making by parties. Indeed, if this data has validity for our field, many mediators may routinely be depriving participants of information they need to consider their options fully, and sacrificing their own instrumental success in helping them to resolve their disputes.



    [i] Symposium, Standards of Professional Conduct in Alternative Dispute Resolution, 1995 J. dispute resolution 95, 108 (1995) (Comments of Lela Love).

    [ii] Today, well conducted meta-analyses of many individual empirical studies help social scientists generalize about empirical findings with greater statistical confidence.  See, generally, Daniel J. O’Keefe, Extracting Dependable Generalizations From the Persuasion Effects Literature: Some Issues in Meta-Analytic Reviews, 58 Communication Monographs 472 (1991).

    [iii] Richard E. Petty, John T. Cacioppo & Martin Heesacker, Effects of Rhetorical Questions on Persuasion: A Cognitive Response Analysis, 40 J. Personality & Social Psych. 432 (1981).

    [iv] John L. Swasy & James M. Munch, Examining the Target of Receiver Elaborations: Rhetorical Question Effects on Source Processing and Persuasion, 11 J. Consumer Res. 877 (1985); James M. Munch & John L. Swasy, Rhetorical Question, Summarization Frequency, and Argument Strength Effects on Recall, 15 J. Consumer Res. 69 (1988).  

    [v]  See, Daniel O’Keefe, How to Handle Opposing Arguments in Persuasive Messages: A Meta-Analytic Review of the Effects of One-Sided and Two-Sided Messages, 22 Communication Yearbook 209-249 (1999) (107 investigations, more than 20,000 participants.)  Another useful meta-analytic review is Mike Allen, Comparing the Persuasive Effectiveness One- and Two-Sided Messages, in Mike Allen & Raymond W. Preiss, Persuasion: Advances Through Meta-Analysis 93-94 (1998). 

    [vi] Daniel O’Keefe, Justification Explicitness and Persuasive Effect: A Meta-Analytic Review of the Effects of Varying Support Articulation in Persuasive Messages, 35 Argument & Advocacy 61 (1998) (23 investigations, 5,358 participants.)

    [vii] Daniel O’Keefe, Standpoint Explicitness and Persuasive Effect: A Meta-Analytic Review of the Effects of Varying Conclusion Articulation in Persuasive Messages, 34 Argument & Advocacy 1 (1997) (32 investigations, 13,754 participants.)

    [viii] See, e.g., Herbert Simons, Persuasion in Society 30 (2001).

    [ix] The basic problem is one of external validity, which concerns the extent to which social science research findings can be generalized beyond the particular study at hand to different people, different settings and different times.  See, generally, Robert Lawless, Jennifer Robbennolt & Thomas Ulen, Empirical Methods in Law 39 (2010). 

    [x] Deborah M. Kolb & Kenneth Kressel, The Realities of Making Talk Work, in WHEN TALK WORKS 483 (1994).

    [xi]  See, e.g., Margaret L. Shaw, Mediator Qualifications: A Report of a Symposium on Critical Issues in Alternative Dispute Resolution, 12 Seton Hall Legis. J. 125 (1988); Chris Honeyman, On Evaluating Mediators, 6 Negot. J. 23 (1990).



James H. Stark is Professor of Law and Director of the Mediation Clinic at the University of Connecticut School of Law.


Douglas N. Frenkel is Morris Shuster Practice Professor of Law and directs the Mediation Clinic at the University of Pennsylvania School of Law.

Their video-integrated text, The Practice of Mediation (Aspen Law and Business 2008), is used widely in law school classrooms. Their article, Changing Minds: The Work of Mediators and Empirical Studies of Persuasion, from which this article is adapted,can be downloaded at




DISPUTE RESOLUTION MAGAZINE is published quarterly (4 times a year) by the American Bar Association Section of Dispute Resolution. Dispute Resolution Magazine provides timely, insightful and resourceful information regarding the latest developments, news and trends in the growing field of dispute resolution throughout the world and features internationally-known scholars and practitioners as authors.


Dispute Resolution Magazine Editorial Board

Joseph B. Stulberg
The Ohio State University Moritz College of Law
Columbus, OH


Andrea Kupfer Schneider
Marquette University Law School
Milwaukee, WI


Chair Emeritus
Frank Sander
Cambridge, MA


James Coben
Hamline University School of Law
St. Paul, MN


Marvin Johnson
Silver Spring, MD


Bennett G. Picker
Stradley Ronon
Philadelphia, PA


Effie D. Silva
McDermott Will & Emory LLP
Miami, FL


Donna Stienstra
Federal Judicial Center
Washington, DC


Frederick Hertz
Oakland, CA


Machteld Pel
Pel Mediation


Council Liaison
Ben Davis
Toledo, OH

Gina Viola Brown


Associate Editor
Louisa Williams

Law Student Editors
Darlene Hemerka

Joyce Fondong

Penn State University

Dickinson Schools of Law


The Editorial Board welcomes the submission of article concepts as well as draft articles relevant to the field of dispute resolution. The Editorial Board reviews all submissions and makes final decisions as to the publication of articles in Dispute Resolution Magazine. Author guidelines for submissions are available below.

Subscription to Dispute Resolution Magazine is included in the membership dues of the Section of Dispute Resolution. Nonmembers of the Dispute Resolution Section may subscribe to the Magazine for $55.00 per year. Additional shipping charges will apply to international addresses. Subscriptions can be purchased online at the ABA web store.  


For more information about becoming a member, see the Member Center.


Section Leadership

Back to Entity's Publications Page

Author Guidelines

  • Advertising with DRM