Recognizing and Addressing Confirmation Bias in Mediation

As defined by the Farnam Street website, confirmation bias is the “tendency to cherry-pick information that confirms our existing beliefs or ideas.” In other words, after making a decision or reaching a conclusion in which we have a vested interest, we tend to interpret any new evidence in a manner that validates our choice, while distinguishing or ignoring facts that contradict or undermine it. The upshot is that confirmation bias clouds our judgment, leading us to continue pursuing erroneous courses of action despite being presented with evidence indicating that our positions are mistaken.

Research suggests that litigators are prone to confirmation bias (which is perhaps unsurprising given the duty of lawyers to zealously advocate for their clients). In a 2010 study of 481 randomly selected lawyers representing parties in civil and criminal cases in the United States (entitled Insightful or Wishful: Lawyers’ Ability to Predict Case Outcomes), researchers asked respondents to identify their minimum goal in a case scheduled for trial, and to estimate the probability of achieving it (on a scale of 0% to 100%). After the cases resolved, the actual outcomes were compared to the predictions.

Researchers found that lawyers who estimated their probability of success at 66% or greater (i.e., those most confident about achieving their minimum goal) erred most frequently in their predictions. In other words, litigators are susceptible to overconfidence.

One key element of the study suggests that lawyer overconfidence stems (at least in part) from confirmation bias. To test whether asking lawyers to play devil’s advocate might reduce overconfidence and result in more accurate predictions, certain lawyers in the study were asked to provide reasons why they might fail to achieve their goal (such as witness and other evidentiary problems, the unpredictability of judges and/or juries, an unappealing client, and the skill of opposing counsel). Surprisingly, contrary to the findings of analogous research in non-legal contexts, researchers found that asking lawyers to consider the weaknesses of their case did not reduce overconfidence rates. In other words, a high percentage of litigators remained convinced they would prevail despite the odds stacked against them.

This finding clearly indicates that litigator overconfidence is traceable at least in part to confirmation bias. Specifically, litigators are prone to ignore factors that reduce the likelihood of success, while interpreting the facts of their case, and applicable law, in an unrealistically optimistic light.

Mediators confirm this anecdotally. For example, Stephen Hochman writes that lawyers “tend to underweight the weaknesses in their case . . . fall in love with their most creative arguments . . . and devalue and reject what they hear from their adversary.” Similarly, my partner David Albalah observes in his seminal article on mediation techniques (entitled For Business Dispute Solutions, Process Matters) that “when business people have ‘lawyered up’ with top-shelf attorneys from prestigious law firms, have invested an exorbitant amount of time and energy into litigation, and have drank their own “Kool-aid,” their legal correctness is cemented in their own mind.”

The authors of the study discussed above noted that overconfidence can lead litigators to recommend rejection of settlement offers their clients would be wise to accept based on a more objective assessment of the merits. This consequence has obvious relevance to the mediation of disputes already in litigation, and it is therefore important for mediators to understand how to identify confirmation bias and develop techniques for dealing with it.

One strategy, adopted by Hochman when making a mediator’s proposal, is to warn both sides as follows:

In the over 350 cases that I previously mediated, there were only seven in which I did not get two yeses to my mediator’s proposal. In six of those cases I got one “yes” and one “no, and in all of those six cases the side that said “no” ended up with a worse litigation or arbitration result than it would have had if it had accepted my proposal.

If both sides trust the mediator’s ability to make an objective assessment of the alternatives to a mediated settlement, this sort of stark warning may persuade lawyers on both sides to accept a mediator’s proposal instead of taking their chances with continued litigation (even if they subjectively believe they could do better than the resolution proposed).

Alternatively, rather than trying to change lawyers’ perception of the likelihood of success in litigation, a mediator can invite the lawyers to wager on the correctness of their views through the use of a high low agreement (which, as previously discussed, represents a “partial” settlement of a dispute under which the parties to a mediation agree to a minimum recovery for the plaintiff, and a maximum payout by the defendant, and then proceed towards final resolution, subject to the agreed upon caps).

For example, the outcome of a case may hinge on whether the court adopts or rejects a certain legal principle. One side may be convinced that the court will adopt the principle, while the other side predicts that the court will decline to do so. Neither side is willing to budge on its conviction. But instead of declaring an impasse, a mediator can invite the parties to bet on their convictions through a high-low agreement that fixes a ceiling on what the defendant will pay even if the plaintiff’s view is vindicated while placing a floor on what the plaintiff will receive even if the defendant’s position prevails. The parties are then free to litigate the relevant legal issue, subject to the high low agreement kicking in once the outcome is known.

Using high low agreements to break impasses at mediation traceable to overconfidence is consistent with academic research recommending the use of “contingent contracts” in analogous business contexts. As their name suggests, such contracts allow parties involved in a negotiation to break an impasse by placing bets on the outcome of a contingent event about which they disagree. In an article entitled Betting on the Future: The Virtues of Contingent Contracts, Professors Max H. Bazerman and James J. Gillespie cite “overconfidence” as a common bias affecting negotiations, and advise readers to use contingent contracts to address it. As they explain:

Companies, like individuals, tend to have an irrational degree of confidence in their own abilities and, as a result, they tend to overestimate the likelihood of achieving positive outcomes. In a contingent contract, each side translates its overconfident assumptions into a wager on the future. The outcome of the wager tends to fall between the two extreme positions, creating a rational result without requiring either party to sacrifice its firmly held bias.

Similarly, in a mediation, to the extent the mediator assesses that confirmation bias is preventing settlement, a high low agreement may provide a solution that gives lawyers the opportunity to prove they are right while mitigating their client’s downside in the event they are wrong.

My partner David Albalah has used high low agreements in this manner to break impasses in insurance-related mediations. In a future post, he will discuss one such case.