61% of plaintiffs who rejected the defendant’s last settlement offer and
proceeded to trial received an equal or worse financial outcome than the offer
they turned down, according to Kiser, Asher and McShane’s 2008 analysis of
2,054 California civil cases — the largest empirical study of settlement
decision-making errors in the published legal literature. The average cost of
that error was $43,100. On the inaction side, the Legal Services Corporation’s
2022 Justice Gap report found that low-income Americans received no or
inadequate legal help for 92% of the civil legal problems that substantially
impacted their lives — leaving the overwhelming majority of valid legal claims
without effective legal action. The data points in opposite directions: the
majority of those who fight tend to lose relative to the settlement on the
table, and the overwhelming majority of substantial civil legal problems among
people without resources go unaddressed.
The asymmetry between plaintiffs and defendants in the Kiser study deserves
attention. Defendants made the wrong decision (rejected an offer that turned
out better than the judgment) only 24% of the time, compared to 61% for
plaintiffs. This gap reflects a systematic pattern: plaintiffs tend to
overestimate the strength of their case and underestimate legal costs, while
defendants — typically insurance carriers or repeat litigants with actuarial
experience — calibrate more accurately. The average defendant error cost was
far higher ($1.14 million) when defendants did lose, but the frequency of
error was much lower. For most individual plaintiffs who are one-time
litigants, the risk is that optimism about court outcomes drives rejection of
settlements that would have served them better.
Both proxies measure something real but neither measures regret directly.
The 61% outcome-error rate captures financial inferiority, not necessarily
felt regret — some plaintiffs pursued litigation for non-financial reasons
(vindication, precedent, principle) and may not regret the decision even
after losing on the financial metric. The 92% no-or-inadequate-help figure
captures the scale of unresolved legal problems among people without resources,
not the subjective experience of wishing they had acted. Case type and jurisdiction
affect the action-side figures substantially: the Kiser dataset draws from
California civil litigation and skews toward personal injury and products
liability, which have higher plaintiff error rates than contract or employment
disputes. Inheritance disputes, tax challenges, and small claims have
structurally different regret dynamics and are not well captured by either
source.
[1]Journal of Empirical Legal Studies (Wiley) — Let's Not Make a Deal: An Empirical Study of Decision Making in Unsuccessful Settlement Negotiations
査読済み
Plaintiffs made the wrong decision in rejecting the defendant's last settlement offer in 61.2% of cases (average cost $43,100); defendants made the wrong decision in 24.3% of cases (average cost $1,140,000)
抜粋
“"Plaintiffs made the wrong decision in rejecting the defendant's last settlement offer or last demand in 61.2 percent of the cases. The average cost of the plaintiff decision error was $43,100. In products liability cases, plaintiffs in 61 percent to 69 percent of cases either recovered nothing or less than the defendant's offer." Table 1 of the same study reports: "plaintiffs committed decision error...in 61.2 percent of the cases. By contrast, defendants committed decision error in 24.3 percent of the cases...there is a substantial difference in mean cost of error between plaintiffs and defendants ($43,100 and $1,140,000, respectively). Given the relatively large discrepancy between the parties' mean cost of error, it is not surprising that the expected cost of error is greater for defendants by a factor of 10."
”
出典データ
2008-09-01
アクセス日
2026-07-03
計算過程
Kiser, Asher & McShane (2008) analyzed 2,054 contested civil cases in California in which both plaintiffs and defendants made settlement demands, rejected offers, and proceeded to arbitration or trial. "Decision error" = plaintiff's final outcome was equal to or worse than the defendant's last settlement offer. The 61.2% plaintiff error rate is used directly as the action-regret proxy: the majority of litigants who chose to fight rather than settle received an inferior financial result. This is not a survey of felt regret but a direct measurement of outcome inferiority — the closest available proxy for ex-post decision regret in civil litigation. Excerpt and statistic updated 2026-07-03 to add the paper's own defendant-side figures (24.3% error rate, $1,140,000 mean cost of error) verbatim from Table 1 of the same study: the body prose cites both the 61% plaintiff figure and the 24%/$1.14M defendant figures, but only the plaintiff figure had been excerpted here. Confirmed against the full published text (Journal of Empirical Legal Studies, Vol. 5, Issue 3, pp. 551-591, September 2008): "defendants committed decision error... in 24.3 percent of the cases" with mean cost of error of "$1,140,000" (p. 566, Table 1).
[1]Legal Services Corporation / NORC at the University of Chicago — The Justice Gap: The Unmet Civil Legal Needs of Low-Income Americans (2022)↗ 1 other entry
政府報告書
Low-income Americans received no or inadequate legal help for 92% of the civil legal problems that substantially impacted their lives in the past year; 74% of low-income households experienced at least one civil legal problem in that year
抜粋
“"Low-income Americans did not receive any legal help or enough legal help for 92% of the problems that substantially impacted their lives in the past year. This study finds that nearly three-quarters (74%) of low-income households have experienced at least one civil legal problem in the past year."
”
出典データ
2022-04-01
アクセス日
2026-05-09
計算過程
Legal Services Corporation contracted NORC at the University of Chicago for a nationally representative survey of more than 5,000 US households at or below 125% of the federal poverty line. The report's two headline figures are distinct: the 74% is a problem-PREVALENCE rate (nearly three-quarters of low-income households experienced at least one civil legal problem in the past year), while the 92% is the UNMET-NEED rate (low-income Americans got no or inadequate legal help for 92% of the problems that substantially impacted their lives). An earlier version of this entry incorrectly read the 74% as "received no legal help whatsoever"; the report makes no such claim. We use the 92% no-or- inadequate-help figure as the inaction-side proxy: it is the closest published measure of how often a substantial civil legal problem goes without effective legal action. It is an access-gap proxy, not a direct regret survey — many of those unmet problems reflect lack of resources or representation rather than a felt decision to walk away, so the rate likely overstates true inaction regret. No direct survey of regret for not pursuing a civil claim exists.
[2]RAND Corporation — The Perception of Justice: Tort Litigants' Views of Trial, Court-Annexed Arbitration, and Judicial Settlement Conferences
信頼できる参考文献
Tort litigants' satisfaction was higher with arbitration and trial than with judicial settlement conferences, suggesting procedural participation has intrinsic value beyond financial outcomes
抜粋
“"The three procedures studied — trial, court-annexed arbitration, and judicial settlement — differed considerably in the procedural fairness and satisfaction ratings they engendered, with arbitration hearings and trials viewed more favorably than settlement conferences. Tort litigants were sensitive to procedural variations, wanting a dignified, careful, and unbiased hearing of their cases."
”
出典データ
1989-01-01
アクセス日
2026-05-09
計算過程
RAND R-3708 surveyed 286 litigants across three state courts. Used here as the corroborating source for the value of legal participation — it demonstrates that people who engage the legal process (even in settlement conferences) report higher procedural satisfaction than those who do not. This supports the inference that non-engagement (inaction) correlates with lower procedural satisfaction and potentially greater regret, though it does not directly measure inaction regret.
注意事項
PROXY MEASUREMENTS ON BOTH SIDES. Neither side uses a direct regret survey. The action rate (0.61) is Kiser et al.'s outcome-error rate from 2,054 California cases: it measures financial outcome inferiority relative to the last settlement offer, not felt regret. Plaintiffs who lost at trial but preferred the process, or who had non-financial goals (vindication, precedent), are counted as "errors" even if they do not regret litigating. The inaction rate (0.92) is the LSC Justice Gap "no or inadequate legal help" figure: the share of substantial civil legal problems among low-income Americans that got no or inadequate legal help — a barrier-to-access measure, not a direct regret measure. (The report's separate 74% figure is a problem-PREVALENCE rate — the share of low-income households that experienced at least one civil legal problem — and does not measure non-engagement or regret; an earlier version of this entry mis-used it.) The 0.92 proxy may substantially overstate actionable regret (much unmet need reflects lack of resources or representation, not a felt choice to walk away) or understate it (those who never knew a legal remedy existed cannot articulate specific regret). The two populations are structurally incomparable: Kiser's litigants had already reached negotiation; LSC's population spans everyone with a substantial civil legal problem, regardless of whether they entered the legal system at all. Case type matters enormously: plaintiff decision errors are highest in products liability (61-69%) and personal injury, lower in contract and employment disputes where plaintiff win rates are higher. In small claims, inheritance disputes, and tax matters, the relevant regret dynamics differ substantially from the California civil litigation sample. The Gilovich inaction-dominance pattern is plausible here — most people who never pursued a valid claim may carry longer-term regret than those who fought and lost — but the data gap on the inaction side means the delta of -0.31 should be read as directional rather than precise.
