Hey, I’m an arbitrator. And like Derek Jeter I always want to improve my game every time I go on the field. This is for me, right?
The topic of the day was “heuristics.” That term means trial-and-error, or a method of learning that is based on cumulative experience. This method, of course, leads to logical error. Knowing a Libyan who wears a moustache does not support the conclusion that all Libyans wear moustaches; and being more familiar with words that start with the letter “k” than with words of which the third letter is “k” does not mean that more letters start with “k” than have it as the third letter. The panelists at the ABA were using these examples of cognitive bias to suggest that decision-makers are prone to certain types of error as a function of these everyday cognitive fallacies.
Well, you can’t argue with that, can you?
Paul Bennett Marrow led off the session by citing five common “heuristics”: availability (the likelihood of something’s occurring in the future is tied to the ease of remembering its happening in the past); overconfidence (individuals overestimate their own chances of success, and reject non-conforming data); hindsight bias (thinking that actual outcomes could have been accurately predicted beforehand); anchoring (the influence of a first value upon the range of subsequent bid-ask negotiation); and representativeness (the categorization of a single datum based on only one of many attributes).
This last “heuristic” was illustrated by former Judge Billie Colombaro, who related a complex tale of unfair firing of a police officer based on the outcome of a polygraph session. Both the policeman’s immediate supervisor and the trial court accepted these results as valid because they were “scientific” or “objective,” and were persuaded that the machine was called a “lie detector” for a reason. Despite a middle-appeals court’s reversal, the state Supreme Court affirmed the trial court’s finding. Judge Colombaro presented this outcome as (I suppose) an instance of collective cognitive bias – a rejection of certain data (i.e., the lack of any basis for concluding that the machine’s recorded responses can result only when the subject is lying) in favor of embracing something that the lie detector results broadly resemble (i.e., reliable objective scientific data such as DNA).
Elizabeth Shampnoi noted that “heuristics” come into play when a matter is ambiguous. That is, we are talking about decision-making in the absence of complete information (which is to say, in my experience, all decision-making). The concern is not so much that the decision-maker is being influenced by cognitive bias, but that the decision-maker may be unaware of that influence. An exercise was performed in which identical damages information was given to the attendees, but half were informed that plaintiff demanded $200,000 and half were told that the demand was $450,000. Mr. Morrow reported that, ordinarily when the exercise was done, the effect of “anchoring” was evident in the increased willingness of the second group to award $150,000.
Really? The sheets of paper in this group were not counted – or if they were counted the count was not reported. When I mediate I see very well that the range of negotiations is often influenced by the opening offer. But am I now being told that my final arbitral award is based on the magnitude of the initial demand? This is not my experience, either as an advocate or as a neutral in arbitration. Indeed, one distinction between litigation and arbitration remains the ability of the parties to select, as the adjudicator, a person whom they trust will decide on the merits and according to industry standards and commercial realities.
ADR practitioners (and litigators for that matter) how are unaware of cognitive bias might do well to bone up on the topic. Dwight Golann’s book is a great place to start, and recent studies on lawyer overconfidence have been rightly noted elsewhere. But the proposition that this is “How Arbitrators Think” remains unsupported by the information that was conveyed in this panel.