A Tale of Two Forums: Employment Discrimination Outcomes in Arbitration and Litigation

Abstract

This article presents data from a novel survey of 1,256 employment plaintiff attorneys to test whether employee rights and remedies are affected by mandatory employment arbitration. By surveying attorneys directly about their most recent employment discrimination cases taken to verdict in arbitration and civil litigation, the author presents a systematic empirical comparison of outcomes between civil courts and arbitration with robust controls. The ability to control for the legal basis of the claim, defendant size, use of summary judgment, and attorney and plaintiff characteristics significantly improves on previous empirical research studies. Consistent with previous research, employee win rates in arbitration are lower than those found in state and federal court. In addition, monetary award amounts and percentage of claim amount awarded to employees who prevail in their cases are significantly lower in arbitration compared to outcomes in state and federal jury trials.

Keywords

arbitration discrimination alternative dispute resolution labor and employment law conflict management

Individual employment rights in the United States have expanded over recent decades through passage of state and federal legislation such as Title VII of the Civil Rights Act of 1964, the Americans with Disabilities Act of 1990, and the Genetic Information Nondiscrimination Act of 2008. These statutes, and the rights they confer, are enforced through a unique system that includes administrative agencies, state and federal courts, and, most recently, private arbitration forums. The effects of enforcing public employment rights in private arbitration are contested in the academic literature, court, and various news outlets. Proponents claim mandatory arbitration is employee-friendly and facilitates access to justice. Critics see an employee-unfriendly forum characterized by inadequate due process and inferior employee remedies. While acknowledging legitimate public policy concerns, the U.S. Supreme Court maintains that mandatory arbitration contracts are broadly enforceable. Indeed, the Supreme Court states in the recent Epic Systems Corp. v. Lewis, 584 U.S. 138 (2018) decision: “Should employees and employers be allowed to agree that any dispute between them will be resolved through one-on-one arbitration? . . . As a matter of policy [this question is] surely debatable. But as a matter of law the answer is clear.”

Recent estimates indicate as many as half of all private-sector employees are subject to mandatory arbitration contracts (Colvin 2017). This finding suggests private-sector workers are nearly 10 times more likely to be covered by mandatory arbitration than by a collective bargaining agreement. Despite the prevalence and controversies surrounding mandatory arbitration, its effects on employee access to justice are not well understood. The academic literature provides some insight about the resolution of employment claims within individual forums such as the American Arbitration Association (AAA) (Bingham 1998; Colvin 2011; Colvin and Gough 2015), the Financial Industry Regulatory Authority (FINRA) (Lamare and Lipsky 2014, 2019; Lamare 2016), federal court (Clermont and Schwab 2004), and state court (Eisenberg and Hill 2003; Oppenheimer 2003). This literature provides only limited controls to account for between-forum variation in case and party characteristics, however. Therefore, it is difficult to determine how, if at all, arbitration affects employees’ rights at work: Is arbitration a genuinely hostile forum for employee claimants, or can simple variation in case characteristics account for between-forum differences in employee outcomes?

The present study addresses this limitation by examining how the forum in which employment rights are adjudicated—arbitration or civil litigation—affects employee remedies under the law. Specifically, this study analyzes data from an original survey of 1,256 employment plaintiff attorneys to investigate the effect of mandatory arbitration on employee win rates and award amounts while accounting for claim, plaintiff, defendant, and attorney characteristics. These data introduce robust controls and offer a unique portrait of employment disputes across several forums beyond what can be found in publicly available data from individual arbitration providers or court dockets. If arbitration is closing the doors to justice, then the mechanism by which the United States has advanced employment rights since the 1960s—individual statutory protections—is being undermined by an institution that systemically favors employers over employees.

Literature Review

Arbitration is a general alternative dispute resolution procedure through which a third-party neutral resolves contractual or statutory disputes. Arbitration is used in a variety of contexts: in employment and commercial disputes; in union and non-union settings; and as a mandatory requirement or voluntary option. This study focuses on one particularly controversial use of arbitration: employer-promulgated pre-dispute mandatory arbitration (hereafter simply “mandatory arbitration”). By signing a mandatory arbitration clause, employees waive their right to sue their employer in court and agree to submit all legal claims to private arbitration. Employers generally present mandatory arbitration clauses to individual non-union employees as “take-it-or-leave-it” contracts of adhesion (Stone 1996; Gough 2016). As in the context of grievance arbitration, awards in mandatory arbitration are binding and subject to limited judicial review. The absence of a union as a countervailing force at contract negotiations, and the coverage of statutory rights,¹ such as Title VII of the Civil Rights Act, however, differentiate mandatory arbitration from grievance arbitration.

Proponents and detractors maintain incompatible perceptions of mandatory arbitration and its effects on the contemporary workplace. Opponents perceive an employee-unfriendly forum and raise concerns over the coercive nature of the contracts (Stone 1996), the inability of employees to access the arbitration forum (Gough 2016; Estlund 2018), lack of diversity among the arbitration profession (Gough and Alpert 2019), biased decision-makers (Colvin and Gough 2015), and inferior employee win rates and award amounts compared to litigation (Colvin 2011; Gough 2014). Proponents of mandatory arbitration concede it is imperfect but assert it is ultimately an employee-friendly forum relative to alternative institutions (Estreicher 2001; Sherwyn, Estreicher, and Heise 2005; Eigen and Sherwyn 2016). They identify an apparent prejudice toward employment disputes in civil litigation and the potential for arbitration to benefit employees through improvements in expediency, cost, and accessibility (Estreicher 2001; Sherwyn et al. 2005; Eigen and Sherwyn 2016).

Empirical scholarship on employment litigation and mandatory arbitration, however, has largely focused on employee outcomes within forums as opposed to systematically comparing employee outcomes between forums (Clermont and Schwab 2004; Colvin 2011; Lamare and Lipksy 2014; Colvin and Gough 2015; Lamare 2016). Indeed, extant empirical scholarship provides a robust picture of how disputes are resolved within individual forums. Yet this same literature offers only relatively crude measures to evaluate which of the above disparate perceptions of mandatory arbitration aligns with reality. To date, comparisons of employee outcomes in litigation and arbitration have relied on simple descriptive statistics that show the population of arbitrated cases receive smaller award amounts and lower employee win rates relative to litigation (Eisenberg and Hill 2003; Colvin 2011; Gough 2014). I explore this literature and its limitations below.

Empirical Evidence on Employee Win Rates

Colvin (2011) analyzed 1,213 mandatory arbitration cases administered between 2001 and 2007 by the American Arbitration Association (AAA), the country’s largest provider of employment arbitration services. He found that employees won 21% of claims taken to verdict in arbitration. In a subsequent study of AAA data through 2014, Colvin and Gough (2015) found that employees won 19% of their arbitration claims proceeding under mandatory arbitration contracts.

With respect to employee win rates in litigation, Eisenberg and Hill (2003) collected large-scale data sets covering multiple forums. Their analysis of 1,430 employment discrimination cases heard in federal courts revealed an employee win rate of 36%. They reported a slightly higher employee win rate of 44% in 160 state court employment discrimination cases. Oppenheimer (2003) reported a 59% employee win rate in a sample of 117 common law discharge cases heard in California State Court. Nielsen, Nelson, and Lancaster (2010) found employees won 33% of their employment discrimination claims in a random sample of 1,672 cases disposed of in federal court between 1988 and 2003. Finally, Clermont and Schwab (2004) accessed the entire universe of federal employment discrimination cases between 1979 and 2000 and reported employee win rates of 38% in cases adjudicated through a jury trial and 20% in judge-only trials. They also noted that judges presided over approximately 12% of trials whereas jury trials were approximately 8 times more common.

Employee win rates provided by Nielsen et al. (2010), Eisenberg and Hill (2003), and Clermont and Schwab (2004) are higher than employee win rates based on mandatory arbitration agreements reported in AAA studies conducted by Colvin (2011) and Colvin and Gough (2015). Rather than clear evidence that mandatory arbitration results in inferior employee win rates, these studies describe win rates within specific forums without providing adequate controls to compare outcomes between forums.

Empirical Evidence on Employee Award Amounts

As it did for employee win rates, empirical evidence suggests employee plaintiffs receive lower award amounts in arbitration than in litigation. Nonetheless, only cautious conclusions can be drawn from these studies. In a study of 1,213 AAA mandatory arbitration cases adjudicated between 2003 and 2007, Colvin (2011) reported a median award amount of $36,500 and a mean award of $109,858. Comparatively, Clermont and Schwab (2004) reported a median award of $70,000 in federal jobs cases reaching verdicts between 1991 and 2001. Eisenberg (2015), however, found that award amounts for employment claims in federal court outpaced inflation every year since 1993; he reported the median damages awarded in federal employment discrimination cases exceeded $300,000 for cases awarded between 2006 and 2012. At the state level, Eisenberg and Hill (2003) found average award amounts exceeded $200,000. Oppenheimer (2003) reported mean damages exceeded $350,000 in 117 California State Court common law discharge cases. Using data from the 2005 Civil Justice Survey of State Courts, the average damages awarded in a random sample of employment verdicts in state courts across the United States is equivalent to $553,000 in 2019 dollars (Cohen 2011).

In sum, these studies reveal monetary damages rendered in arbitration are, on average, substantially smaller than those rendered in civil litigation. Again, using these raw population averages to draw conclusions is difficult in the absence of robust controls.

Interpretive Difficulties

The existing literature reveals stark differences in employee outcomes found in litigation and arbitration. Whether systematic variation exists in the populations of cases being adjudicated between forums is unclear, though. This uncertainty severely limits our ability to draw evaluative conclusions about case outcomes between arbitration and civil litigation. Variation clearly exists in award amounts and win rates between the forums, but there is no natural way to make cases heard in arbitration identical in all respects to cases heard in civil litigation. Can observed variance be attributed to inherent characteristics of the arbitral forum or fundamental differences in the stock and attributes of cases being adjudicated?

Critics of mandatory arbitration argue low employee win rates and damages in arbitration can be explained by inadequate due process protections, rules favoring employers, difficulty obtaining attorney representation, and repeat player effects (Stone 1996; Colvin 2011; Colvin and Gough 2015; Gough 2016; Estlund 2018). Scholars also highlight various benign ways systematic differences may be introduced between the populations of adjudicated cases.

Cases filed in arbitration invariably flow from employers who have self-selected to use the practice. Employers that adopt mandatory arbitration often do so in conjunction with other human resource management and dispute resolution practices (Colvin 2004; Stipanowich and Lamare 2014). Therefore, cases that are ultimately filed in arbitration may have been “filtered” through sophisticated human resource and dispute resolution practices. Lower win rates and awards in arbitration may simply reflect that meritorious disputes are being resolved internally by these larger, more formal employers.

Arbitration is also characterized as more accessible and less expensive than litigation (Estreicher 2001). If it is more accessible than litigation, employees may be more likely to file low-value or riskier cases in arbitration, resulting in the disparate win rates and award amounts described above. Notably, Estreicher (2001) argued that the civil litigation system represents a “Cadillac” and is available only to exceptional individuals with high incomes or high-value claims. The typical employee claimant, he argued, is left with a “rickshaw”; they must proceed on their own and cannot secure attorney representation in litigation because the outcomes are too uncertain or the amount at stake is too small. Completing his analogy, Estreicher (2001) presented arbitration as an accessible and low-cost forum, as a proverbial “Saturn.” As such, arbitration represents a forum in which attorneys are willing to represent “typical” (e.g., low-value or higher risk) cases that would otherwise be excluded from civil court. In such a system, researchers would need to account for this difference in accessibility when comparing outcomes between arbitration and litigation.

Estreicher’s (2001) theory has not been subjected to empirical scrutiny, however. It may also be that the strategic behavior of plaintiff attorneys exacerbates inequalities between litigation and arbitration. If plaintiff attorneys expect lower returns for cases taken to arbitration, as strategic decision-makers, they may be more demanding in their case selection practices. Indeed, lawyers may only agree to represent clients in arbitration if they have exceptional claims or exceptionally large potential damages (Sternlight 2015; Gough 2016; Estlund 2018). The quality of legal counsel may also explain variability in employee remedies between arbitration and litigation (Lamare 2020). Even if employees subject to mandatory arbitration are able to secure attorney representation, attorney bias against mandatory arbitration clauses may affect the quality of counsel employees can access (Gough 2016). If experienced lawyers are less likely to take mandatory arbitration cases because of lower potential payoffs, employees who proceed under such clauses may access lower quality counsel, on average. As Estreicher (2001) suggested, arbitration may change the incentive structure for attorney case selection, but rather than democratize access to attorneys, arbitration may restrict access to representation.

Some scholars contend that the availability of summary judgment, a process whereby judges are requested to rule on a case before a full trial, biases the comparisons of employee outcomes between arbitration and civil litigation (Sherwyn et al. 2005). If summary judgement is common in litigation but not in arbitration, cases dismissed at summary judgement for lack of merit never proceed to a full trial, do not appear in litigation trial outcome statistics, and make comparisons of win rates between the two forums difficult. Indeed, more than half—57%—of federal employment discrimination claims subject to a motion for summary judgment do not survive it (Nielsen et al. 2010). However, emphasizing the role of summary judgement may be misguided. First, this logic does not account for the strategic behavior of parties that settle cases before or after motions of summary judgment who, for example, could use surviving summary judgement to induce settlement of meritorious claims (Nielsen et al. 2010). Perhaps more important, recent research shows motions for summary judgment are filed frequently in arbitration cases, albeit at lower rates relative to litigation (Colvin and Pike 2014; Gough 2014).

Beyond the many plausible avenues introducing variation between arbitrated and litigated cases, a policy-capturing study by Klaas, Mahony, and Wheeler (2006) suggested employment arbitrators are less employee-friendly than other decision-makers. Presenting 32 hypothetical termination decisions to 140 AAA employment arbitrators, 82 labor arbitrators from the National Academy of Arbitrators (NAA), and 83 jurors who had served in employment discrimination cases, they found that labor arbitrators are most likely to rule in favor of employees, followed by jurors, and then employment arbitrators. This finding suggests that employee outcomes are likely to be affected by the forum in which their claims are adjudicated through differences in decision-makers.

I am not aware of any study that directly compares the decision-making of employment arbitrators and judges. Studies suggest, however, that judge-only trials are relatively rare in litigation (Clermont and Schwab 2004). Some empirical evidence supports the conventional wisdom that juries are more favorable decision-makers to employee claimants than are judges (Oppenheimer 2003; Clermont and Schwab 2004). Despite differences in simple descriptive statistics, several quasi-experiments have found more similarities than differences between the decision-making of judges and juries (Robbennolt 2006).

Data and Methods

Given clear differences in outcomes and lingering comparability concerns between cases adjudicated in litigation and arbitration, what effect, if any, does substituting civil litigation for mandatory arbitration have on employee outcomes? This article addresses this question by presenting results from a survey of attorney members of the National Employment Lawyers Association (NELA) and its largest state affiliate, the California Employment Lawyers Association (CELA), about their most recent employment discrimination cases adjudicated in both arbitration and litigation. Founded in 1985, NELA is the largest organization of practicing plaintiff-side employment attorneys in the country with 69 affiliate organizations at the circuit, state, and local levels. Full membership requires attorneys to certify that a majority of their legal practice that involves employment discrimination, illegal workplace harassment, wrongful termination, denial of employee benefits, and other employment-related matters is on behalf of employees.

The survey frame consisted of 1,890 NELA and an additional 828 CELA members. NELA and CELA generously provided contact information for their membership and sent an initial introductory e-mail. Survey solicitation e-mails containing a Qualtrics link were first e-mailed to attorney members. Physical copies of the survey were subsequently mailed to attorneys who had not yet responded. Data collection ended in 2015 after 1,256 responses were received, representing a response rate of more than 46%. Of these responses, 69% were collected from the web-based survey instrument and the remaining 31% were collected via the mailed survey instrument.

NELA and CELA membership were asked to record the following information relating to their most recent employment discrimination cases taken to verdict in arbitration and litigation:² award amount, winning party, employee salary, alleged discriminatory action, whether a motion for summary judgment was filed, defendant size, plaintiff characteristics, among other control variables. Responses were restricted to employment discrimination claims because they are the most common type of employment case filed in federal court and in arbitration (Clermont and Schwab 2004; Colvin 2012; USC 2018). In total, I received information on 556 employment discrimination cases reaching verdict in state or federal courts and 277 employment discrimination cases adjudicated pursuant to mandatory arbitration clauses. An investigation did not reveal noteworthy differences between attorneys who did and did not provide a recent case description. I exclude seven cases that involved voluntary or individually negotiated arbitration clauses to keep the focus of this analysis on mandatory employment arbitration.

These data afford a sample of claims across multiple states, forums, and arbitration providers beyond what the limited public disclosure statements analyzed in previous research have provided (Colvin 2011; Lamare and Lipsky 2014; Colvin and Gough 2015). Though this method allows for the capture of previously unavailable data, I modeled the survey instrument after the Equal Employment Opportunity Commission’s Intake Questionnaire and conducted extensive pretesting of the survey instrument (EEOC 2013). This approach ensured respondents would be familiar with and accurately recall the requested information. Because the survey was anonymous, I could not verify responses by comparing them to court records or published case information.

Surveying NELA and CELA members means that respondent characteristics and experiences may not be representative of the general population of plaintiff-side employment lawyers. This limitation is especially noteworthy because NELA has played a prominent role in legislative efforts to restrict the use of mandatory arbitration agreements (NELA 2017). However, a Federal Judicial Center (2009) study revealed striking similarities between the sample used herein and plaintiff attorney members of the American Bar Association (ABA) Section of Litigation. ABA attorneys and NELA/CELA respondents are virtually identical in terms of years of experience, reliance on contingency fees, percentage of caseload filed in arbitration, and attitudes toward arbitration.

Studies have shown lawyers are prone to overconfidence bias (Goodman-Delahunty, Hartwig, Granhag, and Loftus 2010), and surveys on sensitive topics are generally susceptible to the threat of social desirability bias (Fowler 1988). To mitigate social desirability and overconfidence biases, respondents were promised anonymity and I relied almost exclusively on questions with discrete, objective answers. The absolute win rates and award amounts reported in this analysis (see Table 1), however, are higher than the population averages reported in the most comprehensive studies of arbitration and litigation. Despite being instructed to record their most recent case taken to verdict, attorney respondents likely over-reported cases they won or those they won with large awards. This bias is present in both the arbitrated and litigated data.

Table 1.

Employee Win Rates and Awards

	Litigation	Arbitration
Employee win rate (%)**	62	46
Award amount mean (median)**	$676,688 ($225,000)	$362,390 ($174,000)
% of Claim awarded mean (median)**	82 (75)	55 (43)

Notes: Within-category differences significance levels: ⁺p < .10; *p < .05; **p < .01. ***p < .001.

Dependent Variables

I analyze three dependent variables representing two distinct aspects of adjudicated decisions: employee win rates and award amounts. Although the concept of an employee “win” can be debated,³ for the purpose of this study an employee win (employee win = 1) is any case that plaintiff counsel indicated was adjudicated in favor of the employee claimant. Conversely, an employee loss (employee win = 0) is attributed to any case that plaintiff counsel indicated was adjudicated in favor of defendant employers. The second dependent variable, award amount, is a continuous variable containing monetary award amounts, if any, rendered to successful employee plaintiffs.⁴ A third variable, percentage of claim amount awarded, is a continuous variable calculated by dividing award amount by the initial claim amount, when reported. This is a relative measure to complement award amount that also mitigates certain concerns involving the definition of an employee win.

Independent Variables

Forum, a key variable of interest in this analysis, records the forum in which the employment discrimination case was tried and disposed, including arbitration, federal court–jury trial, federal court–bench trial, state court–jury trial, and state court–bench trial. Of the reported cases, 35% were arbitrated and the remaining were litigated in federal court in a judge-only bench (2%) or a jury (27%) trial or in state court in a judge-only bench (5%) or jury (31%) trial. This ratio of jury to bench trials is similar to those found in previous studies (Clermont and Schwab 2004; Nielsen et al. 2010). The variable filing state specifies where the discrimination charge originated. Records show 33% of cases were filed in California, 6% of cases were filed in New York, 6% were filed in Texas, and the remaining 55% were filed in the other 47 states or the District of Columbia. By including CELA member responses in addition to those of NELA members, California lawyers are over-represented in the sample. To address this concern, I weight all multivariate models using the inverse probability of selection by state (using the proportion of lawyers by state 1) in the survey frame and 2) as published by the American Bar Association).

The legal bases for claims are recorded in a cluster of binary variables based on charge categories developed by the Equal Employment Opportunity Commission (EEOC 2013). The variables sex, disability, race, religion, age, national origin, family medical leave, sexual orientation, and retaliation are coded as 1 if the variable’s namesake was given as a legal basis for the discrimination claim and 0 otherwise. Because more than one legal basis can be given per claim, these variables are not mutually exclusive. The distribution of alleged basis for discrimination is nearly identical between the forums with the exception of age; age discrimination was alleged in 21% of arbitrated but only 14% of litigated cases. The variable summary judgment is coded as 1 if a claim survived a motion for summary judgment prior to final adjudication, and 0 if a motion for summary judgment was not filed.

Employer size is a categorical variable identifying if fewer than 100 employees worked for the defendant at the establishment level, if between 100 and 499 were employed, if 500 to 999 workers were employed, or 1,000 or more workers were employed. Defendants employed fewer than 100 workers in 31% of litigated cases and 23% of arbitrated cases and employed 1,000 or more workers in 39% of litigated cases and 44% of cases adjudicated in arbitration, both statistically significant differences (p < .05). This variable was designed to be a proxy for resources dedicated to defense and formality of internal personnel procedures.⁵Female plaintiff is another binary variable coded as 0 if the plaintiff was male and 1 if the plaintiff was female.⁶ Cases involving a female plaintiff represent 58% of those heard in civil litigation and 52% in arbitration, a marginally significant difference (p < .10). High salary is coded as 0 if plaintiff income was reported to be below $100,000 and 1 if reported to be equal to or greater than $100,000 a year. Employee plaintiffs are 15 percentage points more likely to have high salaries in arbitration than in litigation, a statistically significant difference (p < .01). Claimants had salaries of at least $100,000 in 31% and 16% of arbitrated and litigated cases, respectively. Plaintiff race is another categorical variable coded as either Caucasian, African American, Asian, Hispanic, or other. In litigation, 63% of claimants were Caucasian, 20% were African American, 3% were Asian, 11% were Hispanic, and 3% were listed as other. In arbitration, 58% of claimants were Caucasian, 17% were African American, 8% were Asian, 12% were Hispanic, and in 5% of cases race was recorded as other.

Finally, I have included three measures of representation quality: plaintiff-side specialization, experience, and hourly rate. Plaintiff-side specialization measures the proportion of an attorney’s practice dedicated to representing employee plaintiff claims. Experience is a continuous variable containing attorney responses to the question, “How many years have you practiced employment law?” Likewise, hourly rate is a continuous variable created using attorney responses to the question, “What is your usual hourly rate?” On average, attorneys who reported a claim adjudicated in arbitration dedicate 92% of their practice to plaintiff-side representation, have 20.2 years of experience practicing employment law, and charge an hourly rate of $416 whereas attorneys who reported a claim adjudicated in civil litigation dedicate 94% of their practice to plaintiff-side representation, have 19.3 years of experience, and bill $396 an hour. These differences are of little practical effect but are significant at the p < .05, p < .10, and p < .01 level, respectively.

Results

Table 1 compares average plaintiff win rates and monetary award amounts by forum. Table 2 and Table 3 are expanded tables that provide further descriptive summary statistics. Of the 277 reported cases tried in arbitration, 128, or 46%, were decided in favor of the employee claimant and 149, or 54%, were decided in favor of the employer defendant. Of the 525 reported cases tried in civil litigation, 326, or 62%, were in favor of the employee claimant and the remaining 199, or 38%, were in favor of the employer defendant. Of the 128 and 326 cases awarded damages in arbitration and litigation, respectively, the average award amount rendered to successful discrimination plaintiffs is $362,390 in arbitration and $676,688 in litigation. Employee win rates are significantly higher (p < .001) in litigation when compared to arbitration; precisely, employee win rates in litigation are 16 percentage points, or 35%, higher relative to arbitration. Likewise, the average award reported in litigated cases is 87% higher relative to the average award found in arbitration, a difference which is significant at the p < .001 level. Finally, successful claimants are awarded 82% of the amount claimed in litigation and 55% in arbitration. These findings are both practically and statistically significant (p < .001).

Table 2.

Descriptive Statistics

	Litigation			Arbitration
	Employee win rate (%)	Award amount mean (median)	Percentage of claim awarded mean (median)	Employee win rate (%)	Award amount mean (median)	Percentage of claim awarded mean (median)
Overall	62	$676,688 ($225,000)	82 (75)	46	$362,390 ($174,000)	55% (43%)
Filing state
CA	66	$875,100 ($300,000)	75 (60)	44	$410,149 ($185,000)	50 (32)
NY	61	$392,636 ($147,500)	87 (90)	50	$564,167 ($397,500)	80 (70)
TX	56	$565,333 ($200,000)	99 (29)	25	$123,631 ($87,500)	29 (33)
Other	60	$596,597 ($200,000)	83 (85)	50	$319,067 ($150,000)	57 (50)
Summary judgment
Yes	60	$695,515 ($250,000)	79 (70)	47	$289,452 ($172,000)	51 (36)
No	68	$630,955 ($144,000)	90 (100)	46	$428,033 ($187,500)	58 (50)
Alleged basis for claim
Sex	64	$680,688 ($200,000)	71 (60)	43	$394,646 ($150,000)	58 (45)
Disability	61	$516,059 ($200,000)	80 (75)	50	$326,389 ($167,500)	40 (35)
Race	55	$719,450 ($300,000)	86 (80)	55	$229,417 ($174,000)	43 (28)
Religion	36	$2.66M ($2.66M)	67 (67)	57	$156,000 ($156,000)	41 (41)
Age	53	$584,574 ($323,000)	74 (68)	36	$622,421 ($350,000)	52 (43)
National origin	42	$498,333 ($433,500)	90 (75)	65	$245,085 ($205,000)	29 (28)
FMLA	62	$558,167 ($171,000)	70 (63)	48	$112,091 ($117,000)	63 (69)
Sexual orientation	71	$748,000 ($250,000)	88 (100)	67	$514,000 ($165,500)	160 (160)
Retaliation	67	$789,793 ($330,000)	92 (92)	45	$280,769 ($175,000)	59 (48)
Plaintiff gender
Male	62	$781,102 ($300,000)	77 (73)	49	$393,929 ($192,500)	47 (33)
Female	62	$603,747 ($175,000)	85 (78)	43	$329,724 ($172,000)	63 (60)
Plaintiff race
African American	57	$685,746 ($217,500)	86 (93)	52	$251,001 ($172,000)	66 (38)
Asian	65	$625,583 ($237,500)	155 (188)	52	$428,050 ($275,000)	49 (32)
Caucasian	63	$744,114 ($230,000)	77 (71)	46	$433,906 ($175,000)	59 (53)
Hispanic	64	$444,548 ($212,000)	80 (80)	24	$258,875 ($192,500)	37 (30)
Other	39	$154,714 ($110,000)	18 (18)	62	$150,625 ($118,500)	19 (19)
Plaintiff salary
< $100,000	63	$557,774 ($200,000)	82% (75%)	48	$250,297 ($150,000)	53 (43)
> $100,001	58	$1.34M ($675,000)	78% (75%)	43	$649,626 ($394,500)	59 (43)
Plaintiff company size
1 to 99	67	$413,305 ($180,000)	83 (68)	51	$333,300 ($160,000)	67 (45)
100 to 499	51	$459,227 ($229,000)	96 (88)	44	$353,182 ($195,000)	48 (36)
500 to 999	62	$911,156 ($447,000)	99 (100)	33	$160,550 ($167,500)	64 (100)
1,000 or more	62	$1,100,135 ($378,500)	63 (55)	48	$419,491 ($187,500)	50 (43)

Notes: FMLA, Family and Medical Leave Act.

Table 3.

Distributions by Forum

	Overall	Arbitration	Litigation
N	783	277	556
	100%	33%	67%
Forum (%)
Arbitration	35	—	—
Federal (Bench)	2	—	—
Federal (Jury)	27	—	—
State (Bench)	5	—	—
State (Jury)	31	—	—
Filing state* (%)
CA	33	41	31
NY	6	5	6
TX	6	6	6
Other	55	48	58
Summary judgment* (%)
Yes	70	48	81
No	30	52	19
Alleged basis for claim (%)
Sex	33	32	34
Disability⁺	21	17	23
Race⁺	19	23	18
Religion	2	3	2
Age*	16	21	14
National origin	6	7	5
FMLA	8	9	8
Sexual orientation	1	2	1
Retaliation	28	28	27
Plaintiff gender ⁺ (%)
Female	56	52	58
Male	44	48	42
Plaintiff race* (%)
African American	19	1%	20
Asian	5	8	3
Caucasian	60	58	63
Hispanic	11	12	11
Other	3	5	3
Plaintiff salary** (%)
Less than $100,000	79	69	84
$100,000 or more	21	31	16
Plaintiff company size (%)
1 to 99	29	23	31
100 to 499	20	21	19
500 to 999	12	12	12
1,000 or more	39	44	39
Lawyer characteristics
Plaintiff-side specialization* (% of practice dedicated to plaintiffs)	93	92	94
Years of experience (mean)⁺	19.61	20.22	19.31
Hourly rate (mean)**	$403	$416	$369

Notes: FMLA, Family and Medical Leave Act.

Within-category significance levels: ⁺p < .10; *p < .05; **p < .01.

The relationship between employee outcomes in litigation and arbitration conveyed in Table 1 and Table 2 is consistent with existing empirical studies: Employee win rates and award amounts are substantially lower in arbitration compared to those found in litigation. As previously discussed, however, it is dubious to attribute such differences to an arbitration forum bias by using raw win rates and award amounts without introducing controls.

As previous scholars have noted, population differences between arbitrated and litigated cases may explain differences in outcomes. Differences in claim, plaintiff, and attorney characteristics are explored in Table 3. Cases proceeding under mandatory arbitration clauses are more likely to be filed in California, less likely to have survived a motion for summary judgement, but generally allege a similar basis for discrimination (with the exception of age discrimination). Mandatory arbitration plaintiffs are more likely to be male, more likely to earn salaries above $100,000 a year, but worked for similar-sized firms. Little practical difference in attorney specialization and years of experience is observed between arbitration and litigation; however, attorneys representing clients in arbitration charge higher hourly fees, on average.

To investigate whether these factors explain the statistical relationships evidenced in Table 1, I estimate multivariate regression models for employee win rates and award amounts, provided in Table 4. Each regression is modeled with robust standard errors and state-level probability weights to account for the over-sampling of California attorneys. Model 1 estimates the effect each independent variable has on employee win rates. As employee win is a dichotomous variable, I employ a logit model. Models 2 and 3 are restricted to cases in which employees won their case. Model 2 estimates a negative binomial regression for the monetary award amount. Negative binomial regressions are appropriate when over-dispersion occurs in the dependent variable, as is the case in award amount. Percentage of claim amount awarded is modeled as a Tobit regression to address truncation whereby the distribution of the residuals is approximately normal.⁷ An additional restriction on model 3 is the exclusion of observations in which the claim amount was reported as $0 or left unreported, resulting in a reduced sample size.⁸

Table 4.

Regression Results

	Model 1: Employee wins (Logit)			Model 2: Award amount (negative binomial)			Model 3: Percentage claim amount awarded (Tobit)
	β	Odds ratio	Robust S.E.	β	Award ratio	Robust S.E.	β	Robust S.E.
Forum characteristics
Forum (ref = Arbitration)
Federal (Bench)	0.485	1.624	0.551	0.185	1.203	0.382	0.180	0.153
Federal (Jury)	0.535*	1.707	0.260	1.109***	3.031	0.570	0.217**	0.094
State (Bench)	1.043*	2.837	0.525	0.421	1.523	0.486	0.199	0.163
State (Jury)	0.900**	2.460	0.261	0.978***	2.659	0.431	0.418**	0.131
Claim characteristics
Summary judgment	−0.215	0.807	0.213	−0.172	0.842	0.133	−0.007	0.084
Alleged basis:
Sex	0.181	1.199	0.250	0.250	1.283	0.246	−0.134	0.139
Disability	0.234	1.263	0.281	−0.306	0.736	0.136	−0.004	0.114
Race	0.187	1.206	0.322	0.072	1.074	0.273	−0.070	0.135
Religion	−0.388	0.678	0.604	0.679	1.973	1.292	0.092	0.197
Age	−0.499⁺	0.607	0.277	−0.010	0.990	0.237	−0.049	0.145
National origin	0.804⁺	2.235	0.485	0.110	1.116	0.385	−0.172	0.175
FMLA	0.133	1.142	0.337	−0.034	0.967	0.233	0.049	0.153
Sexual orientation	0.877	2.404	0.708	1.262**	3.531	1.602	0.408	0.253
Retaliation	0.092	1.097	0.250	0.245	1.278	0.234	0.006	0.097
Defendant characteristics
ER size (ref = 1 to 99)
100 to 499	−0.874*	0.417	0.273	0.631**	1.879	0.375	0.231	0.207
500 to 999	−0.057	0.944	0.305	0.238	1.269	0.259	0.144	0.126
1,000 or more	−0.255	0.775	0.247	0.835***	2.304	0.382	−0.085	0.107
Plaintiff characteristics
Female plaintiff	−0.246	0.782	0.212	−0.157	0.855	0.132	0.219⁺	0.114
High salary	−0.224	0.799	0.250	1.154***	3.170	0.521	0.059	0.086
Race (ref = Caucasian)
African American	0.061	1.062	0.311	−0.229	0.795	0.212	0.048	0.121
Asian	0.353	1.424	0.430	−0.121	0.886	0.226	0.365	0.255
Hispanic	−0.768**	0.464	0.294	−0.064	0.938	0.255	0.060	0.147
Other	−0.962⁺	0.382	0.583	−0.982***	0.375	0.115	−0.266⁺	0.145
Lawyer characteristics
Plaintiff-side specialization (%)	−0.005	0.995	0.008	0.010⁺	1.010	.0056	0.002	0.003
Experience (years)	0.008	1.008	0.012	−0.007	0.993	0.008	0.010**	0.004
Hourly rate ($100s per hour)	0.000	1.001	0.001	0.002**	1.002	0.001	0.000	0.000
Constant	0.386			11.529			0.175
N	783			414			207
Pseudo R-squared	0.115			—			0.198

Notes: Models include robust standard errors (S.E.) clustered by state and state-level probability weights. Additional testing showed the results are robust to the removal of outliers. ER, employer; FMLA, Family and Medical Leave Act.

Within-category differences significance levels: ⁺p < .10; *p < .05; **p < .01. ***p < .001.

The coefficients (β) reported in Table 4, model 1 imply that for every unit increase in an independent variable, the odds an employee will win their case will be multiplied by e^^β. This (more intuitive) transformation is presented as the odds ratio. Significant positive coefficients for federal jury trial, state bench trial, and state jury trial indicate an employee’s odds of winning in these forums are higher relative to their odds in arbitration. The coefficient for federal bench trial is positive, though not statistically significant (likely attributable to small sample size), in line with the other litigation forum coefficients. Precisely, compared to arbitration, an employee’s odds of winning increase by 70.7%⁹ in a federal jury trial, 183.7% in a state judge-only bench trial, and 146.0% in a state jury trial. These effects are significant at the p < .05, p < .05, and p < .01 levels, respectively, and support the notion that employee outcomes in arbitration are inferior to what is achieved through civil litigation.

Control variables for procedural differences, plaintiff gender, plaintiff salary, lawyer characteristics, and basis for alleged discrimination were not found to significantly affect the likelihood of an employee win. The coefficient for Age and National Origin are marginally significant. Compared to their Caucasian counterparts, the odds of Hispanic claimants winning their cases are, on average, 46.4% lower. This effect is significant at the p < .01 level. Finally, when defendant firms employ between 100 and 499 workers, the odds an employee will win decrease by 41.7% relative to firms employing between 1 and 99 workers.

The second model in Table 4 measures how predictor variables affect monetary award amounts in cases in which the employee plaintiff wins. Since the natural log is being modeled, it is again more intuitive to interpret the exponentiated coefficients, presented as award ratios, showing the percentage change a one unit increase in the predictor variable has on the monetary damages awarded. Awards rendered in state and federal jury trials were significantly (p < .001) higher than awards in mandatory employment arbitration. Relative to arbitration, monetary damages awarded to successful employee claimants are 203.1% higher in federal jury trials and 165.9% higher in state jury trials, on average.

Of the remaining independent variables, sexual orientation, high salary, hourly rate, and employer size are significant at or below the (p < .05) level. Awards involving employees with at least six-figure salaries are, on average, 217.0% higher than awards for employees who make less than $100,000. Again, the direction, size, and significance of this effect is expected given that back pay, based on former salaries, makes up the overwhelming majority of monetary awards in employment discrimination cases (Sherwyn et al. 2005). Attorneys who charge higher hourly rates were also found to be positively correlated with monetary awards. Every $100 increase in attorney hourly fees is associated with a 20.0% increase in award amount. Though the absolute number of sexual orientation claims is small (see Tables 2 and 3), claims including these allegations receive award amounts 253.1% higher than those that do not. This particular result, however, is not robust to removal of outliers.

The third model in Table 4 shows a Tobit regression with robust standard errors and state-level probability weights for the percentage of claim amount awarded. It provides a complementary analysis of award amounts found in model 2 and provides insight as to the “quality” of employee wins in the various forums. Consistent with model 2, employees are awarded a larger percentage of their claim amount in state and federal jury trials than in arbitration. The percentage of claim amount awarded to successful employees in federal and state jury trials is 21.7 and 41.8 percentage points higher than in arbitration, respectively. These differences are statistically significant (p < .01) and provide further support that employee outcomes in mandatory arbitration are inferior to what is received in civil litigation.

For every additional year of employment law experience, attorneys recover 1 percentage point more of the amount claimed for their clients, on average. Although seemingly practically insignificant, a lawyer with 20 years of experience (the mean value in the data set) will recover an additional 20 percentage points of amount claimed relative to an attorney without experience practicing employment law. The percentage of claim amount awarded is also marginally higher (p < .10) in cases in which the employee plaintiff is female. As a percentage of claim amount, award amounts are 21.9 percentage points greater for female claimants.

Discussion

The main contribution of this study is to examine the effect of mandatory arbitration on employee outcomes when controlling for claim, plaintiff, and attorney characteristics. Results provide strong evidence that attributes embedded in the arbitral forum systematically disfavor employee claimants. This effect is seen in the starkly inferior outcomes that similarly situated employment discrimination plaintiffs experience in arbitration compared to traditional civil litigation: Employees are less likely to win, they receive smaller awards, and they receive a smaller proportion of claim amounts in arbitration relative to state and federal trials. Previous studies have qualified their findings and acknowledged that selection effects may explain the observed relationships in win rates and monetary award amounts. These alternative explanations, however, hold less currency when controlling for characteristics of the claim, plaintiff, defendant, and plaintiff’s counsel.

Indeed, even though dispute resolution forum remains significant in multivariate analyses, results show that arbitrated and litigated cases do vary along key dimensions. Casting a shadow over assertions of arbitration’s accessibility, the distribution of cases involving minorities is nearly identical between the two forums, although females and employees who make less than $100,000 a year appear less often in arbitration. Consistent with previous literature, summary judgement is less frequent in arbitration, though a motion was filed in almost half of all arbitration cases. Finally, I observe little practical difference between the experience and specialization of plaintiff counsel between forums, though a slight difference between the average hourly fees is present. With the notable exception of summary judgement, where differences do exist, they should be associated with more positive outcomes in arbitration relative to litigation; however, this is not what I found.

When controlling for claim, plaintiff, and attorney characteristics, I observe large, statistically significant forum effects across all models. Arbitration claimants are less likely to win and they receive smaller awards than their counterparts in litigation. When comparing arbitration outcomes to outcomes of bench trials, however, I find weaker evidence of inferior results. I caution against drawing strong conclusions based on lack of significance as only 10% of the sample were bench trials, consistent with litigation population statistics (Clermont and Schwab 2004).

Claim characteristics—surviving summary judgement, the alleged basis for discrimination, and plaintiff company size—generally did not explain variance in employee outcomes. The lack of significance for Summary Judgment may seem curious, but sophisticated parties involved in litigation engage in strategic settlement at various points along the litigation process. The lack of significance may be explained by strong cases settling after summary judgment is denied (Albiston and Nielsen 2007). As these data show, researchers should be cautioned against using survival of summary judgment as a proxy for case merits or likelihood of success at trial. Results show the alleged basis for claims are generally insignificant across all models as well. Claims alleging sexual orientation discrimination represent only 1% of total claims but were still related to higher monetary awards in model 2; however, when outliers were removed, this variable lost its significance. Employees of firms with 100 to 499 employees are less likely to win their cases compared to employees of firms with 1 to 99. Perhaps lower win rates can be attributed to increased formality of personnel policies within firms, yet we would expect to find increasingly negative effects associated with firms in the 500 to 999 and 1,000 or more categories, which is not the case. This relationship is difficult to reconcile and should be explored in future studies. One possible explanation could be that employer size is also a proxy for visibility, making larger employers more vulnerable to institutional pressures, for example, from arbitrators, juries, and judges. Perhaps defendant firms with 100 to 499 employees are large enough to benefit from the formality of their personnel policies but not large enough to increase their visibility. This may also explain the firm size coefficients observed in model 2.

Plaintiff characteristics—plaintiff gender, salary level, and race—similarly explained little variation within the models, with notable exceptions. The salary level of the plaintiff was shown to influence the size of awards. As previously mentioned, award amounts in discrimination cases are calculated damages, based on back-pay calculations using (former) salaries, which explains both the direction and size of the salary effect. This finding is notable because high-salaried plaintiffs are relatively overrepresented in arbitration, yet the mean and median award amounts of arbitrated cases remain significantly below those of the litigated sample. Hispanic claimants are less likely to win their case but appear in roughly equal proportion across the forums.

Measures of attorney experience were also shown to have an effect on case outcomes, yet there was little indication that quality of counsel differed between arbitrated and litigated cases. Although lawyers who charge higher fees are associated with larger awards, this could be explained by selection effects. Lawyers who charge more may accept only larger-value cases, or perhaps lawyers who charge higher rates work in markets with higher-income claimants, which is not controlled for using a binary measure of income. Recall that each year of experience is associated with a 1% increase in the percentage of claim amount awarded. So although those who charge higher fees may obtain larger awards in an absolute sense, those with experience receive a higher proportion of claim amounts. One explanation is that experienced lawyers are simply higher-quality lawyers. Another explanation is that experienced lawyers request more realistic claim amounts.

Limitations of my analysis should be recognized. First, although the NELA is the largest organization of plaintiff-side employment lawyers and the CELA is its largest state affiliate, there is no guarantee that the employment discrimination claims these lawyers reported taking to trial are representative of the universe of tried cases in arbitration and litigation. Notably, this survey frame excludes the experiences of employee claimants who are not represented by attorneys and proceed pro se. If possible, future studies should take a random sample of all cases in the various forums to determine the generalizability of these results. Second, this study analyzes only awarded cases and does not consider cases that were settled or dismissed prior to a full hearing. Third, I focus on distributive outcomes of employment discrimination claims without regard to claimants’ satisfaction or feelings toward procedural justice. Sociologists have documented the brutality of the civil litigation process and the dissatisfaction experienced by both winners and losers in litigation (Berrey, Hoffman and Nielsen 2012). A valuable contribution to literature would be made by comparing arbitration and litigation with respect to the elements of procedural justice.

Conclusion

Clermont and Schwab (2004: 429) have said that employment discrimination plaintiffs trying to vindicate their rights in federal court “have a tough row to hoe.” This analysis of contemporary employment discrimination cases shows that plaintiffs in arbitration may have it even worse. Specifically, employment discrimination plaintiffs are less likely to receive a judgment in their favor compared to similar cases disposed in state and federal court. These negative effects are compounded for arbitration claimants through lower award amounts measured absolutely and as a proportion of claim amount. The negative effects associated with arbitration are robust even when controlling for procedural differences between the forums, claim types, as well as defendant, claimant, and counsel characteristics.

From an employee’s perspective, rather than a mere forum substitution, arbitration appears to represent a forced sacrifice of employment rights through lower success rates at trial and lower award amounts when compared to civil litigation. Lower employee win rates and award amounts would incentivize employers and human resource practitioners to adopt arbitration procedures to reduce their exposure to damages under statutory employment rights (Gough 2018). From a public policy perspective, however, inferior results experienced by employee plaintiffs belie arbitration’s reputation as an accessible and adequate substitute for litigation.

I caution against interpreting these findings as a wholesale endorsement of the civil litigation system. Certainly, as the court and supporters of arbitration suggest, with or without arbitration, the doors of the courtroom remain shut to many, if not most, employees with work-related grievances. Employees still must traverse their insecurities about publicly classifying themselves as a victim, public agencies that are out of reach or ineffectual, securing legal representation, and apparent hostility to employment claims—especially those alleging discrimination—from judges and juries themselves (Clermont and Schwab 2004; Nielsen et al. 2010). Even when employee plaintiffs are granted access to their day in court, proceed through trial, and receive a ruling in their favor, as Berrey et al. (2012: 4) reported, employees lament “the institutional barriers they faced in securing competent legal assistance, the devastating toll of litigation on their financial and emotional well-being, and the lack of a clear resolution to their original workplace grievance.”

The ascension of individual employment rights has fundamentally altered the industrial relations landscape and brought the importance of underexplored channels for dispute resolution into sharp relief. These institutions—state and federal courts and arbitration—should not be understood simply as substitutes as the Supreme Court asserts in Gilmer (1991). Rather, as this study suggests, by diluting the remedies available to employee plaintiffs, mandatory arbitration dilutes employees’ rights under the law.

Footnotes

For information regarding the data and/or computer programs used for this study, please address correspondence to the author at mdg228@psu.edu.

ORCID iD

Mark Gough

1

The 14 Penn Plaza v. Pyett, 556 U.S. 247 (2009) decision held that when a union contract “clearly and unmistakably” requires statutory claims to be arbitrated, union employees are foreclosed from pursuing their statutory claims in court. To date, however, unions and employers have yet to widely embrace such language.

2

Respondents were first asked, “Within the last 5 years, have you taken an employment discrimination claim to verdict/award in private arbitration (e.g. AAA)? (labor arbitration does not apply).” If respondents answered “Yes,” they were given the following instructions: “Please describe your most recent employment discrimination case taken to verdict/award through private arbitration, even if you feel the most recent case is atypical or not representative. If you cannot recall precise facts or figures, your best recollection should be recorded.” Respondents were presented with similar branching questions for cases adjudicated through civil litigation.

3

For example, a plaintiff can receive a ruling in their favor without being awarded damages. Additionally, a plaintiff claiming $500,000 in damages can hardly be said to have “won” a case if awarded $1.00.

4

This metric does not account for nonmonetary or injunctive relief. Such relief is rare as reported by Clermont and Schwab (2004), who showed it is given in less than 2% of federal cases, and , who reported injunctive relief is given in less than 3% of cases settled by federal magistrate judges in Chicago.

5

I did not ask about formalization of personnel and evaluation procedures directly, realizing employment attorneys are ill-equipped to answer questions relating to specific internal defendant procedures during pilot testing. Rather, I exploit the well-established positive relationship between size and formalization reported in the literature.

6

The binary construction of this variable, and other plaintiff characteristic variables, is appropriate because the small number of class action cases and those involving multiple plaintiffs were excluded from analysis.

7

This award ratio could also have been modeled using the log of percentage of claim amount awarded (see ). As a robustness check, models were run using a logged measure and results are similar to those reported in this study.

8

The proportion of zero-dollar and missing-claim amounts in the present analysis is consistent with previous studies of AAA (Colvin 2011) and Federal Court (). As a robustness check, I ran model 2 on the sample restricted to the cases used in model 3 and found the results to be materially equivalent.

9

e^(β) − 1 = e^(.535) – 1 = 1.707 – 1 = .707 or 70.7%.

References

Albiston

Catherine R.

Nielsen

Laura B

. 2007. The procedural attack on civil rights: The empirical reality of Buckhannon for public interest litigation. UCLA Law Review 54: 1087–134.

Berrey

Ellen

Hoffman

Steven

Nielsen

Laura B

. 2012. Situated justice: Plaintiffs’ and defendants’ perceptions of fairness in employment civil rights cases. Law and Society Review 46: 1–36.

Bingham

Lisa B.

1998. On repeat players, adhesive contracts, and the use of statistics in judicial review of employment arbitration awards. McGeorge Law Review 29(2): 223–59.

Choi

Stephen

Eisenberg

Theodore

. 2010. Punitive damages in securities arbitration: An empirical study. Cornell Law Faculty Publications Paper No. 391. Ithaca, NY: Cornell University Law School.

Clermont

Kevin

Schwab

Stewart

. 2004. How employment discrimination plaintiffs fare in federal court. Cornell Law Faculty Publications Paper No. 1. Ithaca, NY: Cornell University Law School.

Cohen

Thomas H.

2011. Punitive damage awards in state court, 2005. Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics. Accessed at http://www.bjs.gov/index.cfm?ty=pbdetail&iid=2376 (June 1, 2019).

Colvin

Alexander J. S.

2004. Adoption and use of dispute resolution procedures in the nonunion workplace. In Lewin

David

Kaufman

Bruce E.

(Eds.), Advances in Industrial and Labor Relations, Vol. 13, pp. 69–95. Bingley, UK: Emerald Group Publishing.

Colvin

Alexander J. S

. 2011. An empirical study of employment arbitration: Case outcomes and processes. Journal of Empirical Legal Studies 8(1): 1–23.

Colvin

Alexander J. S

. 2012. American workplace dispute resolution in the individual rights era. International Journal of Human Resource Management 23(3): 459–75.

10.

Colvin

Alexander J. S

. 2017. The growing use of mandatory arbitration. Report to the Economic Policy Institute. Accessed at https://www.epi.org/publication/the-growing-use-of-mandatory-arbitration/ (June 1, 2019).

11.

Colvin

Alexander J. S.

Gough

Mark

. 2015. Individual employment rights arbitration in the United States: Actors and outcomes. ILR Review 68(5): 1019–42.

12.

Colvin

Alexander J. S.

Pike

Kelly

. 2014. Saturns and rickshaws revisited: What kind of employment arbitration system has developed? Ohio State Journal on Dispute Resolution 29(1): 59–84.

13.

[EEOC] Equal Employment Opportunity Commission. 2013. Uniform intake questionnaire. Accessed at http://www.eeoc.gov/form/upload/Uniform-Intake-Questionnaire.pdf (June 1, 2018).

14.

Eigen

Zev

Sherwyn

David

. 2016. Deferring for justice: How administrative agencies can solve the employment dispute quagmire by endorsing an improved arbitration system. Cornell Journal of Law and Public Policy 26(2): 217–73.

15.

Eisenberg

Theodore

. 2015. Four decades of federal civil rights litigation. Journal of Empirical Legal Studies 12(1): 4–28.

16.

Eisenberg

Theodore

Hill

Elizabeth

. 2003. Arbitration and litigation of employment claims: An empirical comparison. Dispute Resolution Journal 58(4): 44–54.

17.

Estlund

Cynthia

. 2018. The black hole of mandatory arbitration. North Carolina Law Review 96: 679–709.

18.

Estreicher

Samuel

. 2001. Saturns for rickshaws: The stakes in the debate over predispute employment arbitration decisions. Ohio State Journal of Dispute Resolution 16(3): 559–70.

19.

Federal Judicial Center. 2009. ABA section of litigation member survey on civil practice: Full report. Accessed at https://www.americanbar.org/content/dam/aba/migrated/litigation/survey/docs/report_aba_report.authcheckdam.pdf (June 1, 2018).

20.

Fowler

Floyd

. 1988. Survey Research Methods. New York: Sage Publications.

21.

Goodman-Delahunty

Jane

Hartwig

Maria

Granhag

Par A.

Loftus

Elizabeth F

. 2010. Insightful or wishful: Lawyers’ ability to predict case outcomes. Psychology, Public Policy, and Law 16(2):133–57.

22.

Gough

Mark

. 2014. The high costs of an inexpensive forum: An empirical analysis of employment discrimination claims heard in arbitration and civil litigation. Berkeley Journal of Employment and Labor Law 35(2): 91–112.

23.

Gough

Mark

. 2016. Employment lawyers and mandatory employment arbitration: Facilitating or forestalling access to justice. Advances in Industrial Relations 16(4): 133–57.

24.

Gough

Mark

. 2018. How do organizational environments and mandatory arbitration shape employment attorney case selection? Evidence from an experimental vignette. Industrial Relations 57(4): 541–67.

25.

Gough

Mark

Albert

Kyle

. 2019. Who supports professional certification? Insights from employment arbitration. British Journal of Industrial Relations 57(4): 850–69.

26.

Klaas

Brian S.

Mahony

Douglas

Wheeler

Hoyt N

. 2006. Decision-making about workplace disputes: A policy-capturing study of employment arbitrators, labor arbitrators, and jurors. Industrial Relations 45(1): 68–95.

27.

Kotkin

Minna J.

2007. Outing outcomes: An empirical study of confidential employment discrimination settlements. Washington and Lee Law Review 64: 111–63.

28.

Lamare

J. Ryan

. 2016. Beyond repeat players: Experience and employment arbitration outcomes in the securities industry. Advances in Industrial and Labor Relations 22: 135–60.

29.

Lamare

J. Ryan

. 2020. The devil is in the details: Attorney effects on employment arbitration outcomes. ILR Review 73(2): 456–78.

30.

Lamare

J. Ryan

Lipsky

David B

. 2014. Employment arbitration in the securities industry: Lessons drawn from recent empirical research. Berkeley Journal of Employment and Labor Law 35(1/2): 113–33.

31.

Lamare

J. Ryan

Lipsky

David B

. 2019. Resolving discrimination complaints in employment arbitration: An analysis of the experience in the securities industry. ILR Review 72(1): 158–84.

32.

[NELA] National Employment Lawyers Association. 2017. About us. Accessed at https://www.nela.org/index.cfm?pg=mandarbitration.

33.

Nielsen

Laura B.

Nelson

Robert L.

Lancaster

Ryon

. 2010. Individual justice or collective legal mobilization? Employment discrimination litigation in the post Civil Rights United States. Journal of Empirical Legal Studies 7(2): 175–201.

34.

Oppenheimer

David B.

2003. Verdicts matter: An empirical study of California employment discrimination and wrongful discharge jury verdicts reveals low success rates for women and minorities. U.C. Davis Law Review 37(2): 511–66.

35.

Robbennolt

Jennifer K.

2006. Apologies and settlement levers. Journal of Empirical Legal Studies 3(2): 333–73.

36.

Sherwyn

David

Estreicher

Samuel

Heise

Michael

. 2005. Assessing the case for employment arbitration: A new direction for empirical research. Stanford Law Review 57: 1557–91.

37.

Sternlight

Jean R.

2015. Disarming employees: How American employers are using mandatory arbitration to deprive workers of legal protection. Brooklyn Law Review 80(4): 1309–56.

38.

Stipanowich

Thomas J.

Lamare

J. Ryan

. 2014. Living with ADR: Evolving perceptions and use of mediation, arbitration, and conflict management in Fortune 1000 corporations. Harvard Negotiation Law Review 19(1): 1–68.

39.

Stone

Katherine V. W.

1996. Mandatory arbitration of individual employment rights: The Yellow Dog Contract of the 1990s. Denver University Law Review 73(4): 1017–50.

40.

United States Courts. 2018. Federal judicial caseload statistics 2018. Accessed at http://www.uscourts.gov/statistics-reports/federal-judicial-caseload-statistics-2018 (June 1, 2019).