2nd Key-Data: Bring Objective Science to Mediation
In 2017, after four years of investigation and detailed review of 47 empirical studies, the American Bar Association Dispute Resolution Section’s Task Force report, “Research on Mediator Techniques,” was released. Its findings were perplexing:
“The Task Force’s review of the studies found that none of the categories of mediator actions has clear, uniform effects across the studies—that is, none consistently has negative effects, positive effects, or no effects - on any of the three sets of mediation outcomes.” The outcome categories are: “(1) settlement and related outcomes, (2) disputants’ perceptions and relationships, and (3) attorneys’ perceptions.”
Yet, we know mediation works. Cases are being settled. Professors are teaching mediation skills. Courts are certifying people to mediate, and disputants and their lawyers are relying on them - all with an accepted shroud of confidentiality. How long can all this sanctioned activity continue without scientific proof supporting it?
We Can do More
Yes, there are obstacles. All, however, are solvable, and strides are being taken to do just that. New survey instruments are forthcoming. Increased academic studies, often using student - or actor - disputants and real mediators, are trying to gain insights into the black box of mediation - its mysterious internal functions that are largely hidden from users, and poorly understood by mediators. But the fact remains, as important as this activity may be, until we directly observe a large enough number of mediations, we will continue to wonder what mediators actually do to help settle disputes. Simply put, nothing can substitute for direct observation of real-life mediations by trained researchers yielding scientifically tested, evidence-based reasons for that reliance.
Fortunately, we already have a variety of methods to give us insight into mediation. We have a range of qualitative methods such as surveys and interviews to improve our understanding, but the use of direct observation methods is fairly rare. For those methods we need to look outside the field of mediation to the research of other disciplines to find new, workable, tested research tools. There we will find a tested scientific method - Behavior Analysis (BA) - that can be used to observe, quantify, and correlate mediation participants’ communication behaviors with both forward movement and impasse. With BA’s methodology, we can link anonymized data from demographic and perception studies with BA’s own anonymized data to identify exactly those techniques that correlate with “yes”.
Interview with Ava Abramowitz and Kenneth Webb
Created in the 1970s, BA, through direct observation, has been successfully applied to analyzing a range of interactive skills. While no single research methodology can provide definitive insights into such a complicated activity as mediation, applying BA to it seems a good first step. Mediation is, after all, a form of facilitated negotiation and persuasion, and that is where BA has made its most visible contribution—Negotiation and Persuasion.
What BA does is discover those communication behaviors that correlate with, if not cause, success. Each behavior is closely defined, so there is no overlap with another behavior. Further, each behavior is objectively defined, defusing the need to guess speaker intent. The language speaks for itself and is coded accordingly.
How does BA do that? BA breaks down communication into its atoms – the individual behaviors people typically use. An idea is put on the table. People usually agree or disagree. Someone is confused. They either seek information or give information. Such behaviors, once identified, are the atoms of communication that can be used singularly, in combinations, or in sequences to achieve results.
An example: For any group to reach a decision, someone has to Initiate—that is, put a proposal on the table; everyone has to share enough information to Clarify the proposal so all are talking about the same thing; and the group has to React. These three key behaviors can be broken down into smaller behaviors, as long as each behavior can be separately and clearly delineated from all other behaviors. Moreover, each behavior must be unquestionably objective, so a high enough degree of inter-rater reliability can be achieved. This allows hypotheses to be tested by anyone certified to observe, results to be replicable, and evidence-based education and training to be created, delivered, and evaluated for efficacy and impact.
The concept of BA is familiar to mediators. Mediators, like negotiators, are taught that active listening is key to conveying empathy; that, when they Seek Information, they should then Test Understanding and later Summarize so the speaker feels heard. This cluster of communication behaviors is taught as the Empathy Loop. Wouldn’t it be wonderful if research uncovered other clusters?
Through BA, scholars and skills trainers could test hypotheses and fine-tune mediator training and performance. Here’s one hypothesis: Different behaviors are used in caucus than are used in joint session. Moreover, different behaviors are more effective in caucus than in joint session. Whatever is discovered, mediators could then learn how to use each milieu strategically, not merely by habit or default.
With research methods such as BA, the mediation profession can finally have objective, quantitative, replicable, science-based insights backing it. We would have a profession-wide meta-language so that Mediator A in Country B could share Initiating strategies with Mediator C in Country D, and both would know they are discussing the same phenomenon. Even better, we could correlate communication behaviors with outcomes from forward movement, settlement, and impasse to discover which behaviors are helpful, when, and where, and which ones are not. We could even link participant perceptions with behaviors and discover which ones induce a higher sense of procedural justice, trust, and neutrality.
And we could do more. We could explore communication behaviors stage by stage from preparation through settlement. We could study nuance and style. Do some negotiators succeed by Proposing, Giving Information, and Disagreeing? Do others succeed by Building on the Proposals of Another, Seeking Information, and Supporting? Are there still other effective style models? Do they differ by economic sector, type of dispute, experience of the parties, by culture, or merely by mediator proclivity?
BA is no magic bullet. Qualitative analysis will invariably be necessary. Best if two differently skilled researchers pair up. Mediation is that complex. Whatever the methodology, it will be difficult to identify behaviors that cause outcome. For a while, correlations will have to suffice, but is using research challenges as a reason to maintain “the black box” an acceptable option? With big-data analytics teasing insights and order from noisy data, it needn’t be.
It’s time to bring objective science to mediation.
1st and 2nd endnotes available below.
 Is it that we know or that we believe mediation works? Is settlement attributable more to the mediator or to the disputants’ desire to settle? A study linking participant expectations pre-mediation with participants’ communication behaviors during mediation as well as their post-mediation perceptions might help us answer that question.
 These obstacles are among the most daunting:
- We have no standard definition of success for either the mediator or the resolution of the dispute. (But see Ava J. Abramowitz, Toward a Definition of Success in Mediation, American Bar Association, Section on Dispute Resolution, Dispute Resolution Magazine, Volume 24, Number 4, Summer 2018.)
- There are no professionally agreed-upon definitions and metrics that enable comparisons.
- Nearly all research to date has been qualitative with each researcher developing their own methodology and terms. Comparability cannot result.
- Getting access to live mediations is more than difficult. The process is confidential—and people like it that way.
- Mediation is contextual. Are the skills demanded of a family mediator the same as those used by a construction mediator? Unlikely.
- Finally, mediation is messy, and any resulting data, qualitative or quantitative, will be noisy. Correlating input and outcome will be hard, finding causations harder still.
 The National Center for State Courts with the assistance of the ABA Section on Dispute Resolution (SDR) is working on “Data Elements for Courts to Collect Regarding ADR,” which will help put the demographics and use of mediation in perspective. Resolution Systems Institute, again with the help of the SDR, has issued “Model Mediation Surveys: A Guide to Courts,” which will provide insight into how people feel about the process and its outcomes. And this is just in America. Academics and researchers around the world are working together to discover what mediators and parties can do to make mediation more effective.
 The word "users" is intentionally selected to include not only the lawyers and their clients (often referred to as “the parties,”) but also experts, family members, corporate employees, and other attendees who the mediator and the parties feel would aid in resolving the dispute.
 “Exactly” probably requires tandem observations by both a qualitative and a quantitative analyst.
 See, for example, Neil Rackham, Peter Honey, & Michael J Colbert, Developing Interactive Skills (1971), Neil Rackham and John Carlisle, The Effective Negotiator — Part I: The Behaviour of Successful Negotiators, 2 J. EUR. INDUS. TRAINING 6, 6-11 (1978), and Neil Rackham, SPIN Selling (1st ed. 1988).
 A mediation, by definition, involves a group—mediator(s), two or more disputants, usually two or more lawyers, and often an expert or more. If an implementable settlement is to be reached, each participant has a role to play in its development. Getting to Yes spells out the most used process. Behavior Analysis details the communication behaviors, the communication atoms, each participant uses and correlates the chosen behaviors with impasse or movement or whatever else the researcher wants to explore.