Making the right decision at mediation

A study reveals the factors leading to favorable settlements

Sanford Gage
2009 February

Conventional wisdom suggests that plaintiffs and defendants are equally likely to make erroneous decisions in rejecting a settlement and proceeding to trial. But conventional wisdom is wrong! Let’s Not Make A Deal: An Empirical Study of Decision Making in Unsuccessful Settlement Negotiations by Randall L. Kiser, Martin A. Asher, and Blakeley B. McShane, published in the Journal of Empirical Legal Studies (September 2008, Volume 5, number 3) analyzed the subject. Their study concluded that more than 60 percent of the time the plaintiff would have been better off by accepting the last offer rather than proceeding to trial. The average loss is more than $40,000 per trial – and that does not include out-of-pocket expenses, attorneys’ fees, emotional trauma, lost opportunity costs, or client dissatisfaction. By comparison, the error rate of defendants is less than 25 percent, although the financial consequences of such errors are far more serious.

Because this statistical study is being referenced widely by mediators, it is important to understand the significance of the report. Accordingly, the discussion that follows will analyze the findings and focus on the implications of these error rates. We will look at why the error rate of defendants and insurance companies is much lower but also why it is far more costly to them in the long run. We also will discuss the various factors that might account for such decisions; and provide practical suggestions for improving your results.

The study results

The empirical study analyzed more than 9,000 settlement decisions made in cases over the last 41 years. Because of the scope of the work, it is estimated that it represented cases handled by about 20 percent of all currently practicing California litigation attorneys. A comparison was made between rejected pretrial settlement offers and actual trial results. Sixty-one percent of the time, the plaintiffs obtained an award that was the same or worse than the result the plaintiff would have achieved by accepting the defendant’s pretrial settlement offer. The defendant decision error was only 24 percent, and where there was insurance, the error rate was only 21 percent.

The initial impression is that plaintiffs are unduly optimistic and that defendants are far more realistic in evaluating cases, thus suggesting that plaintiffs should be more willing to accept the defendant’s last offer. However, while the average cost of the “decision error” by the plaintiff was $43,100, the defendant suffered an astounding average adverse outcome in the amount of $1.14 million. Thus the expected cost of error is more than 20 times greater for the defendant than the plaintiff. Again, that figure only represented the lost settlement opportunity and did not take into consideration all the costs of litigation or other adverse consequences.

Best alternative to a negotiated agreement (BATNA)

Settlement decisions primarily are determined by predictions concerning the outcome of the case if it went to trial. Litigants unwilling to accept an adversary’s settlement offer have decided that a trial represents the best alternative to a negotiated agreement. Although the decision may be rational, the error rate suggests that attorneys and their clients are not correctly evaluating this alternative. As the study illustrates, the majority of the time the plaintiff would have been better off if the party had not gone to trial and in 25 percent of the time, the defendant likewise would have benefited from settlement rather than trial.

Implications for plaintiffs from this study

While there are many factors that go into the decision to reject the last settlement offer of a defendant and proceed to trial, an implication from this study suggests that plaintiffs could benefit substantially by a more thoughtful evaluation of those offers.

The following sets forth some factors that might explain the significantly greater decision error rate by plaintiffs compared to defendants and suggests concepts to secure a better outcome.

Case selection

As the following chart (See Pg. 12) demonstrates, the win rate for plaintiffs varies greatly depending upon the type of case involved. Plaintiffs win approximately 60 percent of personal injury, fraud and breach of contract cases. However, the results drop off to about a 30 to 35 percent win rate for premises liability, products liability, non-PI negligence cases and intentional torts. The poorest success rate in the study related to the difficulty in proving medical-malpractice cases, where plaintiffs only won approximately 20 percent of the time.

The chart also illustrates that in cases where plaintiffs have a high win rate, their overall demands much more closely match the outcome than defendants’ last offers. The larger error rate for plaintiffs might be explained in large measure by the fact that the plaintiff simply lost the case, which of course translates into a poor decision error.

Obviously, in every instance where the plaintiff loses the case, the client would have been better off to have settled for the defendant’s last offer, if there was one. This suggests that the plaintiff attorney should look long and hard at the prospects for settlement before committing the client to trial in doubtful or difficult-liability cases. More significantly, it suggests that plaintiffs’ lawyers should scrutinize cases more carefully at the outset to avoid having to try doubtful- liability cases. It also emphasizes that there is generally an inverse relationship between plaintiff decision error rates and win rates. Plaintiff decision error is lowest in cases with high win rates and highest in cases with low win rates.

Client selection

Client selection merges into case selection. Virtually every potential plaintiff can find a lawyer to bring his or her case to trial. Once the case is accepted, the attorney has a general obligation to see it through. As discovery goes forward, the attorney may realize that the biggest problem with the case is the client. Because the client ultimately has the right to decide whether to accept a settlement offer or proceed to trial, it is likely that a number of cases go forward despite the advice of the plaintiff lawyer. Careful screening at the intake stage and as new information is discovered is vital to limit client selection errors.

When a lawyer accepts a case on a contingency basis, the client might be far more willing to gamble on the outcome because the attorney is fronting costs and not charging for the time spent. High plaintiff error rates were associated with cases in which contingency-fee arrangements are common, such as, personal injury (53 percent error rate) and medical malpractice (81 percent error rate), compared to contracts (42 percent error rate).

One possibility suggested in a footnote to the original article was that lawyers retained on a contingency basis often reflected optimistic overconfidence compared to their probability of success; however, I believe it might be more likely that it is the client that is unrealistic. Early recognition of client difficulties or changed circumstances requires thoughtful decisions by the lawyer to reduce the likelihood of proceeding to trial with a losing case.

Client education

A lawyer should ascertain very early in his or her representation just what the client’s goals are. Problems develop for plaintiff lawyers when the client needs a particular sum of money for some purpose unrelated to the case itself. Such a goal might not be obtainable because it is not realistic.

The lawyer also should ferret out the client who is pursuing the litigation for spite. When the primary goal of the client is to punish the opposition by forcing the case to trial, no settlement offer will be acceptable. The same is true where the client is determined to pursue the case based upon a perceived “principle” requiring vindication by the court.

A concomitant problem relates to client evaluation. Care must be taken not to oversell the client at the outset and create unrealistic expectations. Early client education and awareness that the opening demand is a negotiating figure and not the value of the case will lead to better client control. The time for this discussion is not in the midst of the mediation conference. Where the client has not been educated sufficiently concerning the risks inherent in the litigation and the most likely outcome, conflicts arise. This can create the awkward situation where the lawyer is asking the mediator to sell the client on a settlement that should have been considered with the lawyer beforehand.

Overall, defendants have far more ability to determine the cases that proceed to trial. Most insurance policies provide that the company has the right to settle or not. This offers them a large number of cases to pick and choose the ones that go to trial. By following the motto “settle the good cases” and “try the bad ones,” carriers can force plaintiffs to try marginal cases or those with minimum merit or those with unrealistic settlement demands, thus skewing the statistics in their favor.

Lawyer skills

Although the study involved over 7,000 lawyers, that still is a relatively small percentage of the total of all California attorneys. An effort was made to determine if the experience level of attorneys, the law schools from which they graduated or the size of their firms were useful predictors of favorable or adverse decisions. These personal factors did not appear to be of great statistical significance.

However, no comparison was made of the individual track record of the acknowledged top trial lawyers. It is highly likely that this core group has a much better decision outcome because of their greater skill both in case selection and case presentation. It might well be that the overall statistics suffer from the fact that a larger number of inexperienced lawyers are trying cases for the plaintiff than is true for the defendants. In about one-half of the cases, the defendant was represented by insurance counsel. Generally speaking, those cases will be handled by more experienced lawyers who have a firm to guide them.

The extremely low win rate of medical malpractice cases emphasizes this point. Defense lawyers in this field generally are extremely competent and have excellent resources available, including much greater access to expert witnesses. However, when matched up against experienced plaintiffs’ malpractice trial lawyers, it is likely to be a different story. Such attorneys are extremely careful to select only compelling cases for trial, with strong liability, large economic damages and sympathetic facts. My guess is that this group of specialists accounts for much of the win rates in this area of law.

Defense implications

Defendants had a much lower incidence of decision errors. Overall it was about 25 percent and where insurance was involved, the rate was closer to only 20 percent. This lower error rate probably reflects a number of factors. Decision-making generally will pass through many hands and depending upon the seriousness of the case, many layers. The final evaluation often will be the result of a committee process, rather than a single individual. Moreover, the persons involved in the decision-making process inevitably include experienced claims personnel and lawyers so that collectively they should be able to make relatively sound decisions. This is in sharp contrast with the plaintiffs themselves who often are unsophisticated with little or no litigation experience and in many instances, have retained lawyers who have had limited trial experience.

However, defendants can take little comfort from the results of this study. While vindicating their higher success rate in predicting the outcome, they woefully miss the mark when they are wrong. While the mean cost to the plaintiffs of their decision error is about $43,000, the mean cost to the defendant is a whopping $1.14 million. (Apparently when Colossus makes a mistake, it is colossal.) Incidentally, the expected cost of defendant errors has been rising over time.

Thus, the expected error cost for defendants is exponentially greater than the error cost for plaintiffs. This sobering fact suggests that in the area of high-risk cases the defendant is far more risk-seeking than the plaintiffs, who generally are more risk-averse. This also matches the general thinking that it is defendant errors that result in large plaintiff verdicts rather than risk-taking plaintiff attorneys. Most clients would rather take a minimal settlement than gamble on the outcome of a trial. However, when the defendant makes an offer that is trivial in comparison to the potential outcome, it virtually assures a trial.

Defendants who want to avoid such huge mistakes need to look more carefully at large potential cases. As the chart above demonstrates, defendants seriously misvalued contract, fraud and negligence (non PI) cases and got clobbered as a result. Significantly, these are probably cases where insurance was less likely to provide coverage and dictate settlement decisions.

Plaintiffs generally tended to overvalue cases, seeking about 120 to 125 percent of the eventual verdict. As such, they were not missing the mark by a great amount. In contrast, defendants’ average offer was only about 20 to 25 percent of the verdict, explaining why the defendants’ losses were so large when they made a decision error.

Section 998 offers to compromise

A surprising finding of the study was that a Code of Civil Procedure section 998 (hereafter Section 998) offer made either by the plaintiff or the defendant sharply reduces both the decision errors and the cost of such errors. For plaintiffs, decision error was reduced to 41.2 percent and the mean cost of the error reduced to $19,200. For defendants, the serving of a Section 998 offer reduced the decision error to 6.7 percent.

This significant difference suggests that before a party sends out a Section 998 offer, that party has given far more careful evaluation to the case. It also suggests that such offers often are submitted at very realistic figures once it is determined that it is unlikely that they will be accepted.


Mediation can help reduce decision errors. There is statistical support for this proposition. A group of cases involving attorneys who also were mediators was separately studied. The results showed that plaintiff attorney-mediators decision error rate was reduced to 48.5 percent. Perhaps they are better able to look at both sides of the case. The implication from this study is that mediators can help parties to obtain a more realistic view for settlement. It also suggests that mediators should consider a more evaluative style where necessary to help parties make better decisions rather than limiting their services to a more facilitative role.

Bench trials

Interestingly enough, plaintiff decision error in court trials was reduced but defendant decision error increased. This may have been a reflection of the type of cases more likely to be submitted to a judge for decision and the fact that in such cases the parties typically expected a lower result.


As with court trials, plaintiffs’ decision error rate was reduced in arbitration and defendants’ rate was increased. Significantly, both plaintiffs and defendants had a much lower likelihood of “no error” rate in arbitration. This meant that there was a fairly high likelihood of predictability in such trials.

The existence of past and future damage awards

Behavioral economic theory posits that a party is more likely to recover actual losses already sustained compared to a claim for future economic losses. However, the study did not appear to find a cognitive distinction between “current” damages and “future” damages, where both were alleged. Defendants had a slightly higher decision rate error in such cases, perhaps reflecting skepticism over the nature and extent of claimed future losses.

Punitive damage claims

Not surprisingly, decision error rates and the consequences of such errors were greatest where punitive damages were alleged. Defendant’s error rate increased to 46.2 percent where there were current and future economic damages, coupled with a claim of punitive damages. Although the sample included only 24 cases, the mean offer was about $100,000, the mean demand about $1.25 million and the mean cost of error approximately $3.5 million.

The study commented that while some experimental studies show strong agreement in finding punitive intent, “there is no consensus about how much in the way of dollars is necessary to produce appropriate suffering in a defendant.” Other studies challenge the notion that punitive damages are so unpredictable. Nevertheless, the report concludes that whether the amount of such damages can be predicted or not, that defendants displayed seriously diminished predictive capacity in evaluating punitive-damage claims. Their greater error rate and cost of error suggest that defendants fail to make an adequate evaluative adjustment for such claims, to their detriment.

It is likely that insurance companies are quite good at handling routine, cookie-cutter cases, primarily involving liquidated-damage claims because of their vast experience with them. However, complexity appears to be their enemy. It may be difficult for a claims adjuster or defense attorney to suggest that the company pay a large sum of money voluntarily to settle a high-risk case and much easier to let someone else decide – ie., the jury. This seems to be particularly true with governmental entities today. The decision makers can always blame a “wayward jury” far more easily than justify a large settlement to the inquiring news media.


There appear to be many factors that create the decision errors of both plaintiffs and defendants. Some of the systemic problems primarily affecting plaintiffs include inadequate and unequal information about the case; lack of knowledge about actual outcomes; attorney inexperience; one-shot vs. defendants’ repeat player advantage; and clients taking advantage of the contingency fee arrangement to gamble on the outcome. A systemic problem affecting both parties relates to billing issues. The financial incentive of attorneys to pursue litigation probably more seriously affects defendants, and may cause them to litigate cases that should have been settled.

Psychological factors also affect error rates. Poor judgment on the part of clients and insurers in particular can influence such decisions. This includes litigation supposedly based upon principles such as parties who simply want to harass the other side or be vindicated by a jury verdict, or even to discourage similar litigation by others, unrelated to the actual merits of the case.

Client needs sometimes overtake case values. This may be a plaintiff who needs a particular sum in order to deal with a present need that is unrelated to the merits of the litigation or a defendant that would rather risk all than pay an amount that is too onerous. Differences in parties’ risk aversion or risk taking also can account for such errors. This may reflect overconfidence or unrealistic expectations. Sometimes an aggressive attorney’s need to “prove” something or to be a “winner” in the negotiations forces the case to trial.


Looking at the various factors that might enter into the decision-making process and putting them in the proper perspective could assist a party in securing better outcomes. Consider whether any of the systemic or psychological factors are preventing settlement. Be sure that you are fully prepared for mediation, if that is the settlement forum. Educate your client well in advance and have a plan that recognizes both the favorable and the unfavorable factors. Based on the results of the study, decision making can improve by taking into consideration the use of Section 998 offers.

Try not to get “stuck” in the negotiating process. A fair result might be a good result even where it does not appear to be the “ultimate” outcome. For plaintiffs, rethink the matter when you are within 20 percent of the offer.

For defendants, where the case has substantial potential, particularly with large special damages or punitive exposure, take a careful look at the risk of a catastrophic outcome and be sure that you are factoring the risk adequately. Remember that large plaintiff verdicts generally are a result of defendant decision errors rather than shrewd calculations by the adversary.

Finally, mediation before an experienced trial lawyer who is not invested in the outcome and who is willing to provide a “reality” check to both sides should enable parties to secure better results. Not every case needs to be settled, but making better decisions will provide a greater measure of satisfaction for you and your clients over time.


Sanford Gage Sanford Gage

Bio as of February 2009:


Sanford Gage is a leading mediator and a former president of Consumer Attorneys of California and also Trial Lawyer of The Year. Author of Insurance Bad Faith Litigation (Matthew Bender), he has participated in over 40 appellate cases and authored over 200 articles.

Copyright © 2024 by the author.
For reprint permission, contact the publisher: