I. Introduction
Questions about effectiveness in legal representation and legal services have become more pressing in light of the well-documented and ongoing crisis in access to justice in the United States (Rhode Reference Rhode2004; Abel Reference Abel2010; Selbin et al. Reference Selbin, Jeanne, Anthony and Stephen2012; Albiston and Sandefur Reference Albiston and Sandefur2013). Existing studies of legal representation have almost universally concluded that having a lawyer is better for parties than having no representation (Sandefur Reference Sandefur2010, Reference Sandefur2015; Eagly and Shafer Reference Eagly and Shafer2015; Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016b). At the same time, theory and research also suggest that something less than full representation by a lawyer may be effective for some parties in some civil justice settings. Proposed alternatives to full representation include nonlawyer advocatesFootnote 1 (Rhode Reference Rhode1996, Reference Rhode2004; Zorza Reference Zorza2014), self-help (Zorza Reference Zorza2002; Engler Reference Engler2009), and unbundled legal services (Steinberg Reference Steinberg2011). This article examines representation by nonlawyer advocates and asks: How do nonlawyer advocates develop and exercise legal expertise, and how does this relate to their effectiveness?
Nonlawyers are currently permitted to practice in a number of civil justice settings, including administrative and immigration courts, and are promoted by some scholars and policy makers as an access to justice intervention (Zorza Reference Zorza2002; Rhode Reference Rhode2004; Engler Reference Engler2006; Barton Reference Barton2010; Eagly and Shafer Reference Eagly and Shafer2015). Washington State recently created a program to license nonlawyers to practice law under limited circumstances, not including court appearances, with the express purpose of expanding access to justice (Ambrogi Reference Ambrogi2015). Similar programs are under discussion in other states (New York City Bar 2013; Carlson Reference Carlson2015; State Bar of California 2015).
Scholars know very little about the effectiveness of nonlawyers (Sandefur Reference Sandefur2015). This is only the second original empirical study to examine nonlawyer practice in the United States. In fact, despite strong evidence that lawyers are effective as measured by case outcomes, scholars also know little about the mechanisms of lawyer representation, or what lawyers do that actually helps parties in civil litigation (Abel Reference Abel2010; Selbin et al. Reference Selbin, Jeanne, Anthony and Stephen2012; Albiston and Sandefur Reference Albiston and Sandefur2013; Sandefur Reference Sandefur2015). However, new empirical studies and theoretical exploration look beyond case outcomes to explore the factors that relate to effectiveness in legal representation.
A new meta-study of existing research on civil representation suggests that lawyers' impact may be greatest where they help low-status parties navigate procedures that are simple for lawyers, but complex for unrepresented lay people (Sandefur Reference Sandefur2015). In a previous article, which is based in the same court but asks different research questions and uses a smaller subset of the data examined here, we suggest that lawyers' ability to make strategic choices also plays a role in effectiveness (Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016b). Together, these new findings suggest that knowledge of procedural rules and the ability to interact with judges may play a larger role in lawyer effectiveness than has been previously understood (Sandefur Reference Sandefur2015; Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016b). In this context, an unresolved question is whether nonlawyers can gain the functional equivalent of a lawyer's expertise and thus perform effectively in certain legal and civil litigation settings (Kritzer Reference Kritzer1998; Sandefur Reference Sandefur2015). In response, this article uses the theoretical lens of legal expertise to examine behavioral differences between lawyer and nonlawyer representatives. Using our analysis of these behavioral differences, we subsequently examine how these representatives develop expertise and how the development of expertise might relate to effectiveness. Further, we suggest new factors to consider in future research on legal representation and the development of access to justice interventions.
This article is based on one of the broadest and deepest collections of data about civil representation in recent years, including 5,150 unemployment insurance appeal cases filed in a District of Columbia administrative court from 2011 through 2013, as well as interviews with the lawyers and nonlawyers who practice in the court.Footnote 2 This study goes beyond win rates to ask what representatives do in civil litigation that helps their clients, as well as why representatives behave in particular ways (Abel Reference Abel2010; Albiston and Sandefur Reference Albiston and Sandefur2013; Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016b). The study site offers a unique opportunity to examine lawyers and nonlawyers who practice in the same court and represent the same focal party: employers opposing the grant of unemployment benefits. Unlike most access to justice research, this study examines focal parties who are not low-income individuals (Albiston and Sandefur Reference Albiston and Sandefur2013; Sandefur Reference Sandefur2015). As a result, it expands our understanding of how different party types experience representation.
We begin by discussing the theory of strategic expertise in the context of broader scholarship about legal expertise. We then introduce the hypotheses, the site of the study, and this article's methodological approach. We continue by investigating case outcomes and procedural behaviors for unrepresented, nonlawyer represented, and lawyer represented employer parties. The results of the quantitative analysis present an intriguing puzzle: where employers have nonlawyer representation, the nonlawyers appear in a minority of hearings—the most critical moment in the case. But when nonlawyers do show up to hearings, they engage in certain procedural steps and win cases at the same rates as lawyers. To explain this puzzle, we asked the employer representatives, in qualitative interviews, how they learned to do their jobs and about the choices they make in the course of their work.
The combination of existing theories and this study reveals that how expertise is developed affects how it is exercised, and we propose that this relationship has important implications for understanding access to justice interventions. We find that nonlawyers with specialized knowledge can help parties navigate basic procedures and basic substantive law. However, because the nonlawyers in this court are primarily trained through interactions with the judges before whom they practice and do not possess knowledge from independent legal training, they are ill-suited to influence how these same judges ultimately apply substantive or procedural law. This includes challenging judges' rulings in individual cases (case-focused challenges) and advocating for the interpretation or expansion of legal rules (systemic challenges) (Shanahan, Carpenter, and Mark, Reference Shanahan, Carpenter and Mark2016a). We conclude by identifying implications for future research and access to justice interventions.
II. Theoretical Framework and Hypotheses
Scholars studying legal representation in civil justice, whether through empirical study, theoretical development, or both, are working on two different, but related, paths of inquiry. The first path concerns effectiveness, which is typically measured by case outcomes. Outcome-based studies, particularly randomized controlled trials, can tell us whether representation or some other access to justice intervention has the effect of increasing win rates. However, such studies cannot tell us why representation has this effect, let alone whether the intervention can be replicated successfully in another context (Abel Reference Abel2010; Albiston and Sandefur Reference Albiston and Sandefur2013). As a result, scholars are also working on a second path of inquiry that seeks more nuanced and contextualized ways to understand and measure representation effectiveness, namely, through theories of legal expertise (Abel Reference Abel2010; Albiston and Sandefur Reference Albiston and Sandefur2013; Sandefur Reference Sandefur2015). A goal of this article is to add to this growing conversation about how representation operates and the factors that relate to effectiveness. Though this article engages theory and research on representation effectiveness as measured by case outcomes and reports case outcomes from the data, it focuses on questions of legal expertise. We aim to learn more about what representatives do, why and how they do it, in what contexts, and what all this means for clients as well as for broader issues of access to justice and related research. In particular, we hope to shed light on the little-studied role of nonlawyer advocates in civil litigation.
A. Case Outcomes
The vast majority of the existing literature on representation effectiveness, including years of observational studies and recent randomized controlled trials, focuses on case outcomes as a dependent variable with the fact of representation (most often by a lawyer) as the independent variable.Footnote 3 As synthesized and described by Sandefur (Reference Sandefur2015) in a meta-study of existing research on legal representation, studies conducted to date show that having a lawyer is better than having no representation, and that having a lawyer is also better, although by a lesser degree, than nonlawyer representation.Footnote 4 Existing research suggests it is unlikely that a party represented by a nonlawyer will fare better than one with a lawyer (Sandefur Reference Sandefur2015).
B. Legal Expertise
We have empirical evidence that lawyers serve parties more effectively in civil litigation than nonlawyers as measured by case outcomes, but the field does not fully understand why this is true. Although some expectations exist based on prevailing theoretical work, conflicting empirical accounts make it difficult to generalize or unify various strands of existing theory. For example, Kritzer (Reference Kritzer1998) finds that specialist nonlawyers are more effective than generalist lawyers in certain circumstances, while Miller, Keith, and Holmes (Reference Miller, Keith and Holmes2015) find that generalist lawyers are more effective than specialist lawyers who lack certain skills.
Our approach to understanding effectiveness in representation is based on literature exploring these questions through the concept of professional expertise (McGuire Reference McGuire1995; Kritzer Reference Kritzer1998; Sandefur Reference Sandefur2010, Reference Sandefur2015; Miller, Keith, and Holmes Reference Miller, Keith and Holmes2015; Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016b). This article draws in particular on the works of Kritzer (Reference Kritzer1998) and Sandefur (Reference Sandefur2010, Reference Sandefur2015), which, like this study, focus on legal expertise and representative effectiveness in civil litigation. In this literature, the expertise of lawyers is the starting point for analysis. A key question is: To the extent that lawyers provide effective advocacy (the majority of the data on case outcomes suggests they do), are there particular aspects of what lawyers do, and thus aspects of lawyers' expertise, that are more important than others? Further, if researchers can identify the most important elements of legal expertise in a given civil justice context, can nonlawyers gain such expertise and thus provide effective advocacy in that context?
Sandefur offers two categories of lawyer expertise that we draw on here and in previous work: substantive expertise and relational expertise (Sandefur Reference Sandefur2010, Reference Sandefur2015; Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016b). Substantive expertise stems primarily from formal training and includes knowledge of formal law, procedural rules, and legal documents. Relational expertise can be understood as “people knowledge” or an understanding of how to navigate the human relationships that exist in a given legal context (Kritzer Reference Kritzer1998; Sandefur Reference Sandefur2015). We offer a third type of expertise that adds necessary complexity to understanding lawyers' professional expertise: strategic expertise. This concept captures how lawyers make choices in their work by connecting formal rules and training (substantive expertise) with the informal relationships (relational expertise) they navigate in the course of that work (Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016b).Footnote 5 Strategic expertise explains how lawyers connect formal training with situational understanding and supplement it with strategic thinking and judgment as they serve their clients (Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016b).
On the question of the elements of legal expertise that most relate to effectiveness, Sandefur's (Reference Sandefur2015) meta-analysis finds that substantive legal complexity is not the source of lawyers' biggest impact in existing studies of representation. Instead, lawyers' greatest impact occurs when they represent low-status parties in cases that are procedurally complex for a layperson, but not procedurally complex from a lawyer's perspective. This is particularly true in cases that are typically treated “perfunctorily” by courts, such as eviction cases and disability hearings. In these settings, Sandefur (Reference Sandefur2015) suggests that the most important thing lawyers do to help their clients may be to push courts to follow their own procedural rules. Although we agree that lawyers affect outcomes in such cases and that lawyers' knowledge of procedural rules plays an important role, we differ somewhat in how we analyze the mechanisms at play.
Sandefur's (Reference Sandefur2015) work suggests that the mere fact of lawyer representation acts as an endorsement of low-status parties' cases, thereby compelling judges to potentially view the case as more meritorious and treat the case with greater care.Footnote 6 In addition, Sandefur's research suggests that lawyers help courts follow their own rules by virtue of their presence and the fact that lawyers have knowledge of procedural rules. Our theoretical view follows Sandefur's (Reference Sandefur2015) in emphasizing the importance of lawyers' knowledge of procedural rules, but we theorize that there is something beyond a lawyer's mere presence that matters. We suggest that a representative must do something affirmative with his or her knowledge of the rules, such as arguing for the application of a procedural rule, objecting to the admission of documentary evidence or testimony, or raising the issue of who bears the burden of proof. We also theorize that these acts by lawyers implicate their substantive (formal) knowledge about procedural rules as well as their relational skill (ability to communicate with a judge), and that the most skilled (and most effective) lawyers exercise strategic expertise by combining these two areas when advocating for their clients (Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016b).
Others look at legal expertise through a different lens that emphasizes the role of experience: general versus specialized knowledge. Kritzer (Reference Kritzer1998) finds evidence that nonlawyers with specialized knowledge and experience with a given forum or case type can be more effective in some circumstances than generalist lawyers with no prior experience. In apparent contrast, a recent study of immigration courts suggests specialized knowledge is not necessarily a source of effectiveness for immigration lawyers. In this study, generalist attorneys with stronger overall legal skills and experience fared better at trial than low-skilled lawyers with experience and specialized knowledge (Miller, Keith, and Holmes Reference Miller, Keith and Holmes2015).
Thus, the literature supports the notion that experience matters but strongly suggests that experience alone is not enough. Instead, the most successful representatives may be those who learn to tailor their arguments in judge-specific ways (Kritzer Reference Kritzer1998; Haire, Lindquist, and Hartley Reference Haire, Lindquist and Hartley1999; Miller, Keith, and Holmes Reference Miller, Keith and Holmes2015). Previous studies have suggested that such lawyers build judge-specific reputations over time and that a good reputation relates positively to case outcomes (McGuire Reference McGuire1995; Kritzer Reference Kritzer1998; Szmer, Johnson, and Sarver Reference Szmer, Johnson and Sarver2007; Miller, Keith, and Holmes Reference Miller, Keith and Holmes2015). This may be particularly true in cases that are fact driven and turn strongly on credibility (Miller, Keith, and Holmes Reference Miller, Keith and Holmes2015). In fact, Miller, Keith, and Holmes's (2015) study of asylum cases finds that an attorney's record of winning before a particular judge predicts the likelihood of future wins before that same judge.
Our concept of strategic expertise reconciles differing accounts of general versus specialized knowledge and is consistent with prevailing views on the importance of experience. Our previous theoretical and empirical work suggests that in the Kritzer (Reference Kritzer1998) and Miller, Keith, and Holmes (Reference Miller, Keith and Holmes2015) studies, the more effective advocates were likely the more strategic advocates. It is not enough to simply know the law, and it is not enough to simply know how to communicate with courts and judges; an effective representative synthesizes these two areas of expertise and does so in a highly contextualized way, considering the views and preferences of individual judges. This context-specific synthesis is the essence of strategic expertise.Footnote 7
The literature reveals a consensus that expertise, and how it is exercised, matters. It also raises the unanswered question of whether nonlawyers can develop the functional equivalent of a lawyer's expertise. In response, we argue that one path to answering this question is to understand how legal expertise, in all its forms, is developed by lawyers and nonlawyers. For example, following on the idea that lawyers help parties through their knowledge of procedural rules, we argue that how a representative learns procedural rules will affect how and in what contexts she uses those procedural rules to advocate for her clients. Until now, there has been no empirical exploration of such issues and no comparison of how lawyers and nonlawyers working in the same civil justice context develop and exercise expertise.
C. Expectations About Lawyers and Nonlawyers
Turning to expectations about the lawyers and nonlawyers in the sample, we look first to Kritzer's (Reference Kritzer1998) study of unemployment appeals, which suggests that nonlawyers in this setting are relatively low paid and have little to no formal legal training. Similarly, in this setting, we expect that nonlawyers' substantive expertise, to the extent they have it, will be based on experience handling cases rather than formal legal training. Thus, while they may lack substantive expertise gained through formal means, they likely possess relational expertise gained during their time practicing before the court as well as substantive expertise gained through experience. In contrast, lawyers will have substantive expertise gained through formal legal education as well as substantive expertise gained through experience. It is also likely that the lawyers, like those in Kritzer's (Reference Kritzer1998) study, will have some general relational expertise gained through previous interactions with judges in civil litigation, and specialist attorneys with previous practice experience in the court under study will possess judge-specific relational expertise.
We also take into consideration that the site of this study is a relatively formal administrative court,Footnote 8 which may place nonlawyers at a disadvantage compared to lawyers, assuming the latter possess greater expertise in unemployment law, procedure, and how to interact with judges and court staff. Despite evidence that nonlawyers, including those in this study's sample, may have some specialized legal expertise, we anticipate that lawyers will be more effective advocates than nonlawyers as measured by case outcomes. As such, we derive the following hypothesis:
Hypothesis 1: Workers will win more cases when nonlawyers represent employers, as compared to when lawyers represent employers.Footnote 9
The preceding considerations also inform expectations about lawyer and nonlawyer use of procedural behaviors. Our quantitative data capture five separate procedural behaviors that representatives or parties may engage in before or during a hearing. Given that our previous research shows the use of these procedural steps correlates with success for employers,Footnote 10 we expect a lawyer of even the most minimal level of competence, or an experienced nonlawyer, would engage in these procedural steps (Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016b). However, just as we anticipate that lawyers will have better case outcomes across the sample, we also anticipate that lawyers will be more likely to engage in procedural behaviors than nonlawyers. In particular, we expect to see that lawyers will be more likely to appear in a case. This expectation is based on the nature of lawyers' formal training, which emphasizes their professional obligations to their clients. It is also based on the centrality of in-court work to lawyers' “professional jurisdiction” (Sandefur Reference Sandefur2015). We expect that lawyers will be more engaged in litigating a case than nonlawyers given their formal legal training (substantive expertise) in litigation and use of procedures. Thus, we derive the following hypothesis about procedural behaviors:
Hypothesis 2: Employers with lawyers will engage in procedural behaviors more often than those with nonlawyer representatives.
III. Research Design
A. Study Site
To test these hypotheses, we use research methods that capture the nuances of the data and study site. This study is based on the District of Columbia's Office of Administrative Hearings (OAH).Footnote 11 Unemployment insurance appeal hearings feature both pro se parties and represented parties and three different representative types: lawyers, nonlawyer advocates, and clinical law students. In this article, we focus only on lawyers and nonlawyer advocates who represent employers. The cases follow a predictable timeframe, typically result in a hearing, involve relatively narrow legal questions, and have essentially binary outcomes, all of which make for meaningful case comparisons across the data set.
Existing research suggests that the relative formality and adversarial nature of a given court is relevant to understanding legal representation (Monsma and Lempert Reference Monsma and Lempert1992; Kritzer Reference Kritzer1998; Sandefur Reference Sandefur2015). Thus, we note that OAH is a relatively formal administrative court comparable in style and substance to state-level trial courts where the vast majority of Americans interact with the justice system and is more formal than many state administrative courts (Portillo 2014).Footnote 12 That said, this is a study of a single venue and, as a result, the findings are subject to generalizability limitations, a challenge inherent in any study of a particular civil justice setting. Unusually, all appeals from OAH go directly to the District's highest court, the District of Columbia Court of Appeals, which has been active in recent years in overturning OAH decisions and interpreting the District's unemployment statute.
B. The Data
The data set for this article is more comprehensive than most other studies of legal representation as it includes quantitative case-level data and qualitative data from interviews with representatives. The quantitative data capture party and representative procedural behavior and case outcomes in 5,150 unique cases, including all unemployment appeals at OAH from January 2011 to June 2013 where the circumstances of separation from employment were at issue.Footnote 13 This is not a random sample of all available cases but, rather, every such case in the District of Columbia during the study period. The sample does not include the subset of unemployment appeals regarding eligibility and benefit calculation, as these appeals involve a state agency rather than the employer as the opposing party.Footnote 14 In a previous article based on a subset of this data we asked different research questions and did not parse the data by representative type (Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016b). The qualitative data include interviews with the full range of representative types that practice at OAH. The study is also informed by our experience representing claimants in unemployment insurance appeal hearings before the District of Columbia Office of Administrative Hearings. Though we did not conduct formal qualitative observations of these hearings, two of the authors are clinical law professors who supervised law students in these cases over the course of five years and in more than 100 cases combined. This study is possible because of relationships with OAH judges formed through clinic representation (Charn and Selbin Reference Charn and Selbin2013).
1. The Parties
In unemployment appeals, workers seek benefits while their former employers, the other party to the case, oppose the grant of benefits. Workers include individuals from a range of backgrounds and industries, from low-wage workers to corporate managers, but workers are disproportionately low-income, poor, or people of color (US Government Accountability Office 2007; Nichols and Simms Reference Nichols and Simms2012). Employers, the focal party in this study, represent a diversity of small businesses, large corporations, nonprofit organizations, and government agencies in the District. Employers have an incentive to engage in unemployment cases and to contest benefits because the unemployment insurance program requires employer contribution to the program through payroll taxes, which are based in part on the number of former workers who have received unemployment insurance.Footnote 15 Given the socioeconomic gap between workers and employers and the different incentives the parties face in unemployment appeals, workers and employers inevitably make decisions about securing representatives in different ways.
2. The Representatives
The differences between workers and employers are reflected in the relatively higher overall rate of representation for employers. In 50.9 percent of all OAH unemployment appeals in the sample, at least one party is represented, with employers represented at more than double the rate of workers (42 percent vs. 18 percent). The majority of worker representatives are lawyers (86 percent).Footnote 16 In contrast, nonlawyer advocates provide the majority of employer representation (61 percent).
a. Nonlawyer Advocates
At OAH, a court rule allows nonlawyers to appear on behalf of parties in unemployment appeals, specifically those employed by a company “whose usual business includes providing representation in unemployment insurance cases.”Footnote 17 This rule, similar to rules in place in other states, reflects and supports the existence of a human resources outsourcing industry that provides a range of unemployment, tax, personnel management, and other services to companies, including representation at unemployment appeals hearings (Kritzer Reference Kritzer1998). Hiring these firms is an economic choice on the part of employers; the firms, which we call human resources firms (HR firms), may manage an employer's entire unemployment insurance cost-reduction program or only part of it. Employers choose from a menu of services that might include preventive strategies to reduce unemployment taxes, handling correspondence with OAH and the District agency that administers the unemployment insurance program, collecting and submitting evidence, and representation at hearings (Korkki Reference Korkki2012).Footnote 18
The HR firms with business at OAH range from large national corporations to smaller, regional firms. To provide hearing representation, HR firms contract, on a case-by-case basis, with a network of nonlawyers and with some lawyers, who represent parties in unemployment hearings at OAH. We know of at least one small, regional firm with a full-time staff of nonlawyers who both represent the clients in hearings and handle the clients' other unemployment-related needs. We learned through interviews that nonlawyer contractors are paid approximately $100 per hearing. Notably, this is the same figure (not adjusted for inflation) that Kritzer (Reference Kritzer1998) reported in his study of nonlawyer advocates in Wisconsin unemployment appeals.
b. Lawyers
Lawyer representation on the employer side includes those who may appear at OAH only once, such as in-house counsel, and those who appear routinely. The latter group includes private lawyers who practice in unemployment law, those who take cases on a contract basis from HR firms, and those who serve clients through the Employer Advocacy Program (EAP). Authorized by District law and managed by the Chamber of Commerce, EAP provides free attorney representation to employers (Pumar and Mullen Reference Pumar and Mullen2012). Many of the lawyers who routinely represent employers at OAH also practice in other areas of employment law.
3. Legal Framework
OAH hears all unemployment insurance appeals in the District. The cases are heard by administrative law judges (ALJs) and hearings are de novo, which means the ALJ takes evidence in the case and makes factual determinations and conclusions of law without regard to the underlying District agency determination regarding a worker's qualification for benefits.Footnote 19 These appeals involve very limited motions practice and discovery processes. An unemployment insurance hearing begins with the legal presumption that a worker is entitled to benefits, and the employer bears the burden of proving otherwise.Footnote 20 In termination cases where the legal issue is whether the worker committed misconduct leading to the termination, the employer bears the entire burden of proof. In cases where the worker resigned, the employer bears the burden of proving the worker resigned voluntarily (as opposed to being forced out or functionally terminated); then the worker must prove that the resignation was for good cause, as defined by the statute.
As a result of the legal presumptions and burdens of proof, if a worker attends the hearing and the employer does not, the worker wins automatically. Thus, a basic but critical element of a representative's role, and an area that implicates expertise, is making sure the client attends the hearing. This holds true for employers and workers—if a party has an interest in winning the case, the first and most basic strategic move is attending the hearing. Other scholars have noted the importance of appearance in civil justice settings and have found that lawyer representation tends to increase the likelihood that a party will appear (Legomsky Reference Legomsky2011; Eagly and Shafer Reference Eagly and Shafer2015).
As described in our earlier work, hearings follow a typical framework, including testimony, documentary evidence, and closing statements (Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016b). There is also a limited disclosure requirement of providing notice to the court and the opposing party of any documents or witnesses that the party plans to use at the hearing.Footnote 21
C. Data Collection
To identify a set of cases for this article, we collected unemployment benefits appeals hearing data from cases filed in the District of Columbia Office of Administrative Hearings between January 2011 and June 2013. This data set encompasses 5,150 unique cases over the study period. We attempted to code every possible procedural element of each case, though only a subset of these procedures are addressed in this article.Footnote 22 We note that we did not collect party names to protect confidentiality; therefore, we cannot analyze party demographics, such as an employer's industry or size, in this study.
To collect the quantitative case data, we engaged in a three-step process. First, we downloaded data from the court's case-management system. Second, we supplemented and verified these data through review of each paper case file, conducted according to a comprehensive collection protocol. Third, we performed supplemental two-tier data checks of the paper case files and reviewed the collected data for both internal consistency and consistency with court procedures. We then coded the collected data according to a comprehensive coding plan to allow for the use of statistical software for analysis.
To conduct the qualitative interviews, we developed a set of standardized questions to be asked of each representative. The questions covered a range of topics, including how representatives find and select clients, learn to handle unemployment hearings, approach case handling, and use substantive law and procedure, as well as the extent to which representatives interact with one another on individual or organizational levels.
We used the quantitative case data to identify representatives and contacted those representatives who appeared most often before the court, including lawyers, law firms, nonlawyers, and supervising attorneys at law school clinical programs. We conducted semi-structured interviews through telephone conversations, and, where a representative could not speak via telephone for any reason, through written responses to questions. We completed an interview with every representative who consented to the interview. All but three interviews were conducted via telephone with two interviewers on each call: one note-taker and one person to ask questions. Two interviewees submitted written responses to the questions in lieu of a telephone interview.
We interviewed three nonlawyer advocates and two lawyers who represent employers. We also interviewed three lawyers who represent workers and two supervising attorneys from law school clinics, although we do not report those findings here. We know that the final interview subjects make up a significant proportion of the total representation in unemployment cases at OAH. However, to protect confidentiality, we cannot offer details about the proportion of cases handled by each representative. To further protect the confidentiality of interview subjects given the small number of total representatives who practice at OAH, we use female pronouns to identify all representatives in this article. The sample includes male and female representatives.
1. Dependent Variables
The main dependent variables in the quantitative portion of the study are derived from case-level data and include case outcomes (the likelihood of a worker winning) and the procedural behaviors of parties and their representatives. The main dependent variables in the qualitative portion of the study, which are derived from qualitative interviews with employer representatives, are the factors that shape the behavior of employer lawyers and nonlawyers.
To measure the relationship between representation and case outcomes, we have operationalized a “win” as a favorable outcome for a worker (1) and a “loss” as an unfavorable outcome (0). Thus, we will discuss employers' case outcomes in terms of the likelihood of a worker winning. On average, workers win 59 percent of the cases in this sample. To measure procedural behavior of employers and their representatives, the presence of each behavior is noted as 1, the absence of the behavior as 0. Across the sample, employers and their representatives engage in procedural behaviors at the following rates: employer representative appearance at a hearing (occurs in 49 percent of the cases in the sample);Footnote 23 employer party appearance at a hearing (47 percent); disclosure of documents in advance of a hearing (42 percent);Footnote 24 introduction of documents during a hearing (35 percent);Footnote 25 and testimony of an employer witness during a hearing (41 percent).
We look to qualitative data to explain the phenomena observed in quantitative data. In particular, we look for factors that explain differences and similarities between lawyer and nonlawyer procedural behaviors, such as factors that explain why a representative would appear at a hearing or not, or why a representative would present testimony or not.
2. Independent Variables
The type of representative retained by an employer (a lawyer or a nonlawyer) and the balance of representation in a case (by employer and worker representation type) are the two main independent variables in the quantitative analysis. The type of representation retained by an employer is also the main independent variable of interest in the qualitative portion of this study.
To introduce the puzzle presented in this article, we first test the relationship between different forms of representation balance in cases against the likelihood of a worker winning a case. Thus, we look beyond the baseline question of whether a party has representation or not to explore whether the type of representative a party has may also influence outcomes.
In this analysis, we operationalize representation balance through the full range of combinations of the presence and absence of representation for both parties, disaggregated by type of representation. For workers, this means representation type is split between clinical law student representatives and lawyers. For employers, representation type is split between nonlawyer representation and lawyers. These balance combinations are dummied out for ease of interpretation.Footnote 26
3. Control Variables
In the quantitative analyses, we introduce two control variables that may have independent effects on the dependent variables of interest. In the case outcomes model, where the likelihood of a worker winning is the dependent variable, we control for the unobserved characteristics of individual judges' decision-making behavior by grouping cases by the judge hearing the case. In this study, there are sixteen total judges who oversee the hearings in the sample, with an average of 320 cases heard per judge.Footnote 27 The model accounts for each judge's cases by developing a “cluster” of cases per judge. The modeling strategy using these clusters is detailed below.
In the case outcomes model and the procedural behaviors model, we also control for appellant party. If the worker brings the appeal, this variable is denoted as 1, if the employer brings the appeal, this variable is denoted as 0. In 71 percent of the cases, the appellant is the worker, which means that a worker has appealed an agency-level determination denying his or her claim for unemployment benefits.
D. Quantitative Methodology
To demonstrate the relationship between the dependent and independent variables of interest, it is important to make methodological choices that best allow us to test the hypotheses. Random intercept (also known as random effects) logit and binary logistic regression are the most appropriate modeling choices given the data's structure and the dependent variables of interest.
In the first series of models, random intercept logit is the most appropriate because we seek to demonstrate the effect of a categorical independent variable, the balance of representative type, on a binary dependent variable, the case outcome for workers. Because the cases are happening in the same court, but with different judges presiding, it is important to control for “judge-level” effects, or unobserved heterogeneity between cases happening across judges (Bartels Reference Bartels and Franzese2015). The random intercept model accounts for the variance in workers' likelihood of winning across judges.Footnote 28
For the second series of models, we use a binary logistic regression to model the likelihood of employer party and representative engagement in certain procedural behaviors present in the data. Each behavior—employer representative appearance, employer party appearance, disclosure of documents before a hearing, introduction of documents at a hearing, and testimony by an employer witness—is operationalized as a binary indicator. In each model, the retention of a lawyer or nonlawyer is compared to a baseline measure of not retaining a representative. This logistic regression model measures the likelihood of an employer party or representative engaging in particular behaviors, controlling for who brought the appeal.
E. Qualitative Methodology
In addition to the quantitative case data, we have conducted semi-structured interviews, the goal of which was to obtain information that would help us contextualize and understand the statistical results, thus improving the explanatory power of this research. In the process of operationalizing concepts of interest for a quantitative study, we recognized that some of the underlying complexity of the concepts becomes lost in the process. Qualitative interviewing allows us to reinsert some of the nuance present in the concepts back into the empirical analysis itself (Creswell Reference Creswell2014). By interviewing a variety of representative types, we improve our ability to interpret the quantitative data and thus strengthen empirical claims (Creswell and Clark Reference Creswell and Plano Clark2011).
F. Methodological Considerations for an Observational Study
This study is observational in nature. We do not use a randomized design due to the nature of access to the data and the ethical challenges of randomizing representation. However, as an observational study, it does include every unemployment case filed in the court over a two and a half-year period where the reason for a worker's separation from employment is at issue.Footnote 29 We are careful to not call what we do “causal” and instead use statistical methods to compare groups and demonstrate the correlative relationships between representation and case and procedural outcomes.
We also use qualitative data to add layers of context and additional interpretive power to the research, thus strengthening our ability to understand and suggest explanations for the phenomena observed in the quantitative data. Even though this analysis is not operating from a random sample, the approach of using a large sample, collecting extensive data on each case, and adding qualitative information allows us to consider differences and similarities across the data, and thus to explore important insights into the operation of legal representation and to test the theories developed in this article (Abel Reference Abel2010; Albiston and Sandefur Reference Albiston and Sandefur2013). We believe this research adds nuance to the understanding of legal representation and contributes to the scholarly conversation about researching representation and civil justice, as well as the practical conversation around the value and need for given access to justice reforms.
IV. Quantitative Results
A. Case Outcomes
Table 1 presents the results from the first series of models, which examine worker case outcomes, or likelihood of winning, across the full range of balance of representation scenarios that exist in the data. For example, in a given case, there may be no representation on either side (balanced, no representation) or a worker may have no representation while the employer has a nonlawyer (WK no rep, ER nonlawyer). As discussed below, we examine party representation in two ways. First, we split the data by representative type for each party and thus look separately at lawyer and clinic representation for workers (Model 1), and separately at lawyer and nonlawyer representation for employers (Model 2). Second, we combine lawyer and clinic representation into a single variable, which is expressed as “WK any rep” for workers, and combine lawyer and nonlawyer representation into a single variable expressed as “ER any rep” for employers. This modeling is based on the theoretical assumptions that lawyer and clinic student representation will have a similar relationship to workers' case outcomes, while lawyers and nonlawyers will be associated with statistically different outcomes.
In Model 1, workers' likelihood of winning in the baseline (or excluded) category—the category where we expect workers are at the greatest disadvantage (WK no rep, ER lawyer)—is compared to likelihood of winning across every other representation balance scenario.Footnote 30 In this model, workers are better off, or win more hearings, when they are represented and the employer retains no representation, or when the employer is represented by either lawyers or nonlawyers (as a combined measure), as compared to when workers are not represented and the employer retains any representation. Model 1 is consistent with the results from our first article, where we compared unrepresented parties to parties that had any type of representation (but did not parse the data by representative type) (Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016b).Footnote 31
Table 1 Likelihood of Worker Winning Across Balance of Representation
| Model 1 | Model 2 | Model 3 | |
|---|---|---|---|
| Balanced, No Representation (1) | –.034 | .494 | –.035 |
| (.068) | (.096) * | (.068) | |
| WK No Rep, ER Nonlawyer (2) | .840 | ||
| (.107) * | |||
| WK No Rep, ER Lawyer (3) | |||
| WK Any Rep, ER Lawyer (10) | 1.32 | ||
| (.176) * | |||
| WK Any Rep, ER Nonlawyer (11) | 1.49 | ||
| (.175) * | |||
| WK Any Rep, ER No Rep (12) | 1.26 | 1.79 | |
| (.132) * | (.149) * | ||
| WK Any Rep, ER Any Rep(13) | .886 | ||
| (.132) * | |||
| WK Lawyer, ER No Rep (15) | 1.20 | ||
| (.141) * | |||
| WK Clinic, ER No Rep (16) | 1.56 | ||
| (.321) * | |||
| WK Lawyer, ER Any Rep (17) | .931 | ||
| (.128) * | |||
| WK Clinic, ER Any Rep (18) | .580 | ||
| (.298) * | |||
| Appellant Party (1 = Worker) | –.608 | –.606 | –.605 |
| (.069) * | (.070) * | (.069) * | |
| Baseline: WK No Rep, ER Any Rep (M1, M3) or Lawyer (M2) (14) | .773 (.108) * | .252 (.128) * | .771 (.108) * |
| N # obs, # groups (obs per group min/avg/max) | 5,120, 16 (3/320/724) | 5,120, 16 (3/320/724) | 5,120, 16 (3/320/724) |
| Wald Chi2 | 232.27 * | 290.43 * | 233.21 * |
| /lnsig2u | −2.34 | −2.26 | −2.34 |
| (.491) | (.493) | (.490) | |
| sigma_u | .310 | .323 | .310 |
| (.076) | (.079) | (.076) | |
| rho | .028 | .031 | .028 |
| (.014) | (.014) | (.014) | |
| lr test of rho = 0 | 82.99 * | 86.73 * | 82.72 * |
| Likelihood ratio test (M2 vs. M1) | 62.84 * | ||
| Likelihood ratio test (M3 vs. M1) | 2.43 |
Notes: Logistic regression with fixed effects produced comparable results. We note the problematic nature of using fixed effects in nonlinear models (Bartels Reference Bartels and Franzese2015). Statistical significance denoted by
* =p<.05. WK = worker. ER = employer.
Based on our theoretical assumptions, we then tested the relative model fit between Models 1 and 2. Model 2 is an extension of Model 1 and separates out the representation categories for employer representatives. In comparing these two models, we find that Model 2 has an improved fit over Model 1 and that cases where employers have lawyers show worse outcomes for workers as compared to cases in which employers have nonlawyer representation. We do not find the same pattern when splitting off worker attorneys and clinics (in comparing the model fit in Model 3 vs. Model 1), suggesting the empirics confirm our theoretical assumptions: nonlawyers are different from lawyers, while clinical students and lawyers are not different when controlling for judge-level effects and appellant party status. Further, we tested the equality of the coefficients of employers with lawyers or nonlawyers when workers were unrepresented. We found that in Model 2 (M2), where the groups are separated, the coefficients are significantly different between the two groups (p = .004). This is further indication that the types of representation for employers should be separated. The coefficients for the groups in M3—splitting up clinics and lawyers for workers—were not significantly different from one another (p = .28, .26). We conclude from these tests, as well as the likelihood ratio tests, that the properly specified model is M2.
The estimated effect of representation imbalances in Model 2 is interpreted graphically in Figure 1. We find evidence supporting our expectation that in situations where representation imbalance favors employers, and lawyers as compared to nonlawyers represent them, employers will be more likely to win against unrepresented workers. In our baseline category, where unrepresented workers face lawyer represented employers (Category 14), workers are predicted to win 47.6 percent of these cases, all else equal. However, when nonlawyers represent employers (Category 2), the predicted probability of an unrepresented worker winning increases to 67.5 percent, all else equal.Footnote 32 What could explain the high likelihood of winning for unrepresented workers when they face nonlawyer represented employers, particularly in light of the high rates of nonlawyer representation?
1 Predicted Probabilities for Workers Winning Based on Representation Status.
Our theoretical approach led to the initial hypothesis that lawyers and nonlawyers would have different relationships to case outcomes, with nonlawyer representation correlating with a higher likelihood of winning for workers (and, thus, worse outcomes for employers) as compared to lawyer representation. The data in the worker case outcomes models are consistent with the hypothesis. Further, the wide gulf in predicted outcomes in cases between the representation types employers may select in these cases lead us to question why there are differences between lawyers and nonlawyers. Specifically, is there something that lawyers and nonlawyers are doing differently in their procedural behaviors that explains these differences in case outcomes?
B. Procedural Behaviors
If the data reveal more about a worker's likelihood of winning a case when considering nonlawyers and lawyers separately on the employer side, it should be meaningful to understand more about the differences in how lawyers and nonlawyers behave before and during hearings. Table 2 presents the results from the models of different procedural behaviors employers and their representatives may engage in before or during a hearing.
Table 2 Employer Behavior by Representation Type
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | |
| Employer has lawyer | 2.91 | 2.15 | 2.59 | 2.10 | 1.91 |
| (.124)* | (.101)* | (.104)* | (.090)* | (.091)* | |
| Predicted probability (ER has lawyer) | 72.7% | 75.8% | 83.4% | 83.1% | |
| Employer has nonlawyer | .615 | 1.13 | .813 | .581 | |
| (.069)* | (.074)* | (.073)* | (.071)* | ||
| Predicted probability (ER has nonlawyer) | 42.9% | 46.4% | 52.8% | 55.7% | |
| Employer is appellee | .283 | –.557 | −1.15 | –.426 | –.568 |
| (.110)* | (.069)* | (.074)* | (.071)* | (.068)* | |
| Constant (excluded category) | −1.09 | –.186 | –.237 | –.861 | –.409 |
| (.089)* | (.067)* | (.070)* | (.072)* | (.067)* | |
| Chi2 | 546.81* | 568.81* | 965.75* | 659.93* | 568.67* |
| Pseudo R 2 | 0.272 | 0.105 | 0.192 | 0.112 | 0.095 |
Notes: Models: (1) representative appearance; (2) party appearance; (3) document disclosure; (4) introduce documents; (5) witness testimony.
In Model 1 of Table 2, we compare the appearance rates of representatives in hearings where an employer retains representation and find that lawyers are significantly more likely to appear at a hearing than nonlawyers (in Model 1, the excluded category). In Models 2–5, we examine whether employers with lawyer or nonlawyer representation are more likely to engage in certain behaviors present in the data set (employer party appearance at the hearing, disclosure of documents prior to a hearing, introduction and admission of documents at a hearing, presentation of employer witness testimony) compared to employers who are not represented.
We find that employers with representation in a case are more likely to engage in all of these behaviors as compared to when they are not represented, and that employers with lawyers are significantly more likely to engage in them than employers with nonlawyer representation. For example, when lawyers represent employers, the predicted probability of introducing documents is almost 73 percent, as compared to only 43 percent when nonlawyers represent employers. This trend continues throughout the other behaviors we studied, with lawyer represented employers having higher predicted probabilities of presenting testimony (75.8 percent as compared to 46.4 percent), ensuring client appearance at the hearing (83.4 percent as compared to 52.8 percent), and disclosing documents (83 percent as compared to 56 percent).
These findings confirm the second hypothesis that lawyers and nonlawyers engage in procedural behaviors differently and are supported by the case outcome results discussed earlier. Lawyers are more likely than nonlawyers to appear at a hearing, to have their client appear, to disclose documents and witnesses in advance of a hearing, to introduce documents at a hearing, and to present testimony. These results are consistent with the hypothesis about a representative's procedural behaviors, with the theoretical expectations of lawyers' greater expertise in case handling, with the expectations regarding incentives to appear at hearings, and with our earlier research in this court (Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016b).
C. The Appearance and Procedural Behavior Puzzles
It is no surprise that all of the behaviors employers and their representatives may engage in during a hearing are highly correlated with a representative's appearance at a hearing.Footnote 33 As noted earlier, in unemployment appeals litigation, the hearing is the determinative moment in the case. There is no discovery process or pre-trial motion practice, the hearings are de novo, and, perhaps most importantly, the employer bears the burden of proof. Thus, if an employer does not appear, it cannot engage in any procedural behaviors and the worker is almost guaranteed to win the case. In addition, we previously found that having a representative who strategically engages in procedural behaviors correlates with higher win rates for the represented party (Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016b).
Notably, when appearance is controlled for in either the case outcomes models or the procedural behavior models, the differences between case outcomes and procedural behaviors used by lawyers and nonlawyers disappear. Given the importance of appearance as a baseline matter and the connection between strategic usage of procedural behaviors and successful hearings, two puzzles remain: What explains the discrepancy in appearance rates for employer lawyers (88 percent) and third-party representatives (28 percent)? And what explains the similarity in procedural behaviors and case outcomes when employers and nonlawyers appear at hearings? Put simply, why do nonlawyers look so different from lawyers in the aggregate, but so similar in the 28 percent of cases where nonlawyers actually appear at a hearing? Our qualitative data help solve these puzzles.
V. Discussion and Qualitative Results
In this section, we use results from qualitative interviews to explore the appearance and procedural behavior puzzles. First, we explain how the appearance puzzle relates to representatives' expertise. Next, we discuss the different ways that lawyers and nonlawyers develop and exercise legal expertise, with the notable finding that judges play a powerful role in shaping nonlawyer expertise, and explore how the development of expertise (how representatives learn) affects how expertise is exercised (what representatives ultimately do in their work).
A. The Appearance Puzzle
Previous research shows that a basic but critical function lawyers serve is getting their clients to show up for court (Larson Reference Larson2006). Indeed, if helping clients navigate procedural complexity is a key domain of lawyers' effectiveness, a result supported by the literature, ensuring client appearance at court is among the most basic but most important procedural steps. While the importance of showing up for court may seem obvious, the reality of civil litigation involves a high level of default, particularly on the part of low-income defendants (Larson Reference Larson2006). In this court, employers are the better-resourced party, but non-appearance essentially guarantees a loss for employers because they bear the burden of proof.Footnote 34 Representative appearance is also strongly correlated with party appearance (see the Appendix). Finally, most employers who have a representative of record will not receive service of process from the court or opposing party—service will generally be sent only to the representative. Given the importance of appearance in unemployment appeals, how can it be that lawyer and nonlawyer appearance rates are so drastically different (88 percent vs. 28 percent)?
The qualitative interviews demonstrate that nonlawyer behavior, in the aggregate, is so different from lawyers' behavior because these two representative types work in different institutional contexts that shape and limit the exercise of expertise. In fact, while individual nonlawyer representatives have strong economic incentives to show up at hearings, as they are paid on a per-hearing basis, they have no control over the initial choice to attend a hearing. They simply cannot exercise any expertise in the context of this critical choice. In contrast, lawyers do have control over the choice to attend a hearing, as well as strong economic and professional incentives to attend, and the data suggest that lawyers routinely advise their clients to appear at hearings, a choice that nonlawyers cannot make for institutional reasons. Thus, when an employer chooses a representative, either a lawyer or nonlawyer, the employer is (perhaps unknowingly) making a decision about the extent to which the hearing representative (the person with legal expertise in the OAH hearing context) may exercise expertise on the employer's behalf.
Recall that an employer has three choices in unemployment appeals: go pro se; hire a lawyer directly; or hire an HR firm, which may send a nonlawyer or lawyer representative. If an employer chooses a lawyer, that lawyer is the gatekeeper of all decisions related to the representation. However, if the employer hires an HR firm, the gatekeeper is a nonlawyer staff person at the HR firm known as a “service coordinator.” Here, it is important to recall that if an employer has a representative of record, that representative, not the employer, will receive all service related to the case, including the notice of a hearing. Thus, representatives play a powerful gatekeeping role.
The interview subjects report that employers enter into service agreements when they hire HR firms. Some employers demand representation at all hearings and contract with the HR firm to receive this service—the service coordinator simply implements the prearranged service. However, the interview subjects report that for most employers, decisions about hearing representation are made on a case-by-case basis. An HR firm manager states that service contracts “typically” include hearing representation “upon request and if available” (emphasis added), a finding that helps explain the low rate of appearance by nonlawyers.
Turning briefly to economic incentives for HR firms, we note the interview subjects' reluctance to discuss what employers pay for HR firm services (and in the case of nonlawyer advocates, a lack of information about such costs). However, the data strongly suggest that contracts for as-needed representation, where employers pay a flat fee for a package of services that includes hearing representation, are the industry norm. Another possibility is that HR firms charge an “add-on” fee for every hearing, but none of the interview subjects report this practice. If, as we believe, HR firms contract for as-needed representation in the context of a flat fee for a package of services, the firms have a clear economic incentive not to send representatives to hearings.
The notion that economic incentives drive the low rate of nonlawyer appearance is consistent with information provided by a manager at an HR firm who reported that “the industry” generally favors avoiding hearings and appears to be moving away from hearing representation across the country in an effort to cut costs. As this manager stated: “The way to have dirt cheap prices is not to provide hearings [sic] representation.” The manager also suggested that smaller, regional firms were more likely to provide hearing representation, as compared to large, national firms. Thus, the interviews suggest that discouraging representation at hearings in case-by-case decisions is common in the HR firm industry, and likely the norm among big, national firms. From an employer perspective, the HR firm practice of discouraging attendance at hearings is likely enabled by corollary economic incentives for employers. Employers bear the cost of paying their employee witnesses to attend hearings, an endeavor that easily consumes half or more of a workday.
Once the decision to attend a hearing is made, an HR firm's service coordinator contacts the nonlawyer advocate with the hearing date. If the nonlawyer is available and wants the work, she will accept the case without knowing the employer's name or anything else about the case. HR firms also contract with lawyers to represent employers at hearings, and these lawyers, like nonlawyer advocates, simply decide based on availability and have no ongoing relationship with the employer. No one we spoke with knew how the decision to send a lawyer versus a nonlawyer is made or whether it is driven by employers or by HR firms. One possibility is that some employers choose to pay a premium for lawyer representation as part of their menu of services.Footnote 35
Turning to lawyers (other than those who contract with HR firms), we believe the high appearance rate of lawyers is driven by a combination of professional norms and other incentives. The ethical rules that govern the legal profession set a baseline standard of zealousness and diligence. Further, as a general matter, a lawyer who agrees to represent someone in a case and makes the representation known to a court must have a substantive reason for not showing up at the deciding moment of the case (D.C. Rules of Prof. Conduct 1.16). Obvious economic incentives also relate to hearing appearance—a lawyer who does not appear is unlikely to be compensated or to receive repeat business from a client.
Unlike nonlawyer advocates, lawyers have the power to make substantive decisions about which clients to represent. The findings about lawyers' screening practices contradict the common idea that lawyers “cherry pick” cases and screen heavily for perceived merit (Genn Reference Genn1993; Engler Reference Engler2009; Greiner and Pattanayak Reference Greiner and Pattanayak2012; Miller, Keith, and Holmes Reference Miller, Keith and Holmes2015). The lawyers we interviewed state that they advise clients against attending a hearing if they think, as one put it, that a case is “weak or a death trap.” However, lawyers generally express willingness to represent employers at hearings as long as they see “anything resembling a case.” This approach reflects the reality that unemployment law allows for a range of potential outcomes, even in potentially “weak” cases, depending on the other party's appearance or evidence. In contrast, HR firms clearly engage in such cherry picking and actively discourage employers from attending hearings, whether based on concerns about cost or perceived merit.
B. The Procedural Behavior Puzzle
With a greater understanding about why the appearance rates of lawyers and nonlawyers are so drastically different, we ask: Why are case outcomes and procedural behaviors so similar when lawyers and nonlawyers do appear at hearings? To understand these similarities, we use qualitative interviews to gather information about how lawyer and nonlawyer representatives develop their expertise and how this relates to the exercise of expertise. We find that the similar procedural behaviors reflect specialized expertise in unemployment hearings that lawyers and nonlawyers share. However, we also find that this quantitative similarity belies a key difference. Namely, we find that nonlawyers can and do help parties navigate basic, commonly used procedures and that they understand basic substantive legal concepts. However, nonlawyers are not equipped to challenge judges on issues of substantive law or procedure in a given case, or to expand or develop the law in any way. In contrast, lawyers can and do challenge judges on points of substantive law and procedure, and can and do argue for law reform. Next, we explain the source and nature of these differences.
1. Developing Expertise
The quantitative findings and qualitative interviews reveal that nonlawyers can develop sufficient legal expertise in unemployment hearings to help their clients navigate commonly used procedures at OAH, at least at the level captured by the data. The nonlawyers understand basic procedural and evidentiary steps, including how to introduce exhibits (such as authenticating a document), the role of burdens of proof, and the importance of witnesses with personal knowledge of the facts in dispute. They also understand basic substantive legal elements, for example, the factors an employer must prove to show a worker was fired for violating a company policy. In addition, they appear to have skill in communicating with judges and court staff.
Nonlawyer expertise is developed on the job. As one nonlawyer described her training, “I shadowed someone for two hearings, then, trial and error.” Only one of the nonlawyers had prior legal training, as a paralegal; the others had no legal experience prior to beginning the job. In addition, none of the nonlawyers receive any form of training, supervision, or continuing education from the HR firms who employ them. As one stated: “In the industry nothing is done to train or educate independent hearings people. The industry wants to do as little as possible.” Another described her training by saying, “I learned the very hard way. The first company that I worked for, my predecessor took me to a hearing and then I was thrown into it. I was so green it was pathetic. My training was on the job.”
Despite a lack of formal training, all the representatives interviewed have twenty or more years of experience in unemployment appeal hearings.Footnote 36 They described three specific sources as critical to their on-the-job learning: workers' lawyers (the opposing party); other nonlawyer advocates; and, most importantly, judges. Notably, none of the nonlawyers reported reading the statute or regulations that govern unemployment law in the District, and none recall reading the court's procedural rules. Only one reported attempting to follow developments in appellate law, although she noted difficulty in finding cases, and stated that nonlawyer advocates “tend to have little if any knowledge of precedent cases that come out.”
Nonlawyers recalled learning from opposing party lawyers and other nonlawyer advocates. One said, “I would learn a lot from the attorneys representing the workers … oh, that's how we object.” One described learning to use a particular strategic move that involves calling the worker as the first witness in the employer's case-in-chief in a voluntary quit case. When asked how she learned to do this, she credited another nonlawyer advocate with teaching her.
Above all, nonlawyers emphasized the central importance of learning from judges. Some nonlawyers spoke of particular judges, those who have been particularly helpful, in glowing, affectionate terms. One exclaimed, “I love Judge [X]!” Another said, “Judge [Y], who I also think is fabulous, he is the nicest man.” The nonlawyers offered stories about judges teaching them both procedural and substantive law. These lessons were imparted in two contexts: the hearing room and the judges' written decisions. As one nonlawyer stated, she learned through a combination of “the hearing itself, [the] judge during the hearing, and [written] decisions.”
Nonlawyers described learning during hearings by paying attention to how and what kind of decisions judges make, but also through moments of direct teaching, such as a judge telling a nonlawyer how to handle a particular procedure or point of law. A nonlawyer advocate who began her career outside the District recalled one such instance:Footnote 37
I remember when I offered a document [at OAH], as you would in Texas, or Maryland, or Virginia. A judge said, “Authenticate it. Identify the signatures, ask if there have been changes or alteration, where is the document maintained, if it's time records, how did you obtain it?” I think [I learned] from trial and error.
When asked how they learned court procedures, none of the nonlawyers recalled reading OAH's procedural rules. Instead, they point to judges and experience in the hearing room as their sources of knowledge about procedures. Nonlawyers emphasized how they learned procedure directly from judges and also reflected an understanding that procedures vary between judges. For example, one noted, “I try to pay attention to how procedures vary. What one judge wants is not what another judge wants.”
Turning to learning outside of the hearing room, nonlawyers describe how they learned substantive unemployment law and how to meet the evidentiary and legal burden in a case by reading judges' written opinions. One representative recalls how a judge began sending her copies of her written decisions. Another said: “If I can get [written opinions] I read them because this is the way I become better educated. You know what a certain judge wants and what another judge doesn't like.”
Finally, in a statement that captures the menu of sources of nonlawyer expertise, one described her learning in this way:
Definitely from reading final orders. But also judges. Like Judge [X] … [he's] not just explaining to employer witnesses, but is talking to the claimant. He'll say, “This is what I have to look at in this case.” [He will] explain in layman's language. [He is] very kind to both parties, [he says] “I want them to know this is what I have to determine.” Judge [Y] does that [also]. I learned by listening to them [the judges] during hearings, reading, [the] hearing process itself, [and] picking up how certain judges look at certain aspects more carefully.
In contrast to the OAH-specific training of nonlawyers, lawyers developed expertise through formal legal training, previous litigation experience, and, finally, experience at OAH. Like nonlawyers, the lawyers learned through trial and error. However, they also possessed preexisting legal expertise, applied this expertise to the unemployment hearing context, and then supplemented it with OAH experience. Lawyers have the same specialized expertise in unemployment hearings as nonlawyers, but also a broader and deeper legal expertise gained through varied litigation experiences.
Knowledge of substantive law was the starting point for lawyers when they began doing hearings—all report that reading the unemployment statute and case law was the first step in their learning process. As one explained:
The cases started for me when [a] client came to me and said, “I have UI matter.” The first thing I did was check the law, of course, the statute, to familiarize myself with the law. And it's pretty straightforward. So, I went to hearing and represented the client. I kept doing it.
Lawyers' expertise is also shaped by previous litigation experience. For them, OAH is one forum, and unemployment cases one discrete area of law, out of many fora and areas of law where they practice. The lawyers have general civil litigation experience and existing practices focused on employment law. Their experience in courts of general jurisdiction and other administrative courts informs their work in unemployment appeals. As one stated:
As part of our employment law practice, we represent employees at [Equal Employment Opportunity Commission] hearings, so we do these kinds of hearings as part of our business. Fifty percent of the practice is devoted to that type of work. I'm very familiar with the administrative process, submitting information before the hearing, burdens of proof, and just the process of litigation in general.
Lawyers also develop specialized expertise through ongoing experience in unemployment hearings. For example: “Nothing takes the place of sitting first chair. On-the-job training. [You're] thrown different things. [You] sit there and work through it … nothing takes the place of sitting there.”
In sum, lawyers and nonlawyers both develop legal expertise through experience at hearings—through trial and error. However, nonlawyers develop expertise in a narrow, specialized context and from a very specific source (OAH judges), while lawyers draw on a range of sources and contexts to inform their practice. Next we ask: How do these different paths to developing expertise relate to what representatives actually do in their work?
2. Exercising Expertise
Turning to how representatives exercise expertise, we first note areas of similarity in lawyer and nonlawyer expertise. Nonlawyers developed, through experience, specialized expertise in basic legal principles and the basic procedure of unemployment hearings. For example, nonlawyers understand employers have the burden of proof, the importance of bringing witnesses with personal knowledge of the facts at issue, and the appropriate language for authenticating a document. Such legal concepts and processes would be challenging for most laypeople, but they are a part of nonlawyers' expertise and an area where they undoubtedly help their clients. Through context-specific training in unemployment hearings, nonlawyers develop the functional equivalent of certain aspects of lawyers' legal expertise and can help their clients navigate fundamentals of the hearing process and unemployment law. This qualitative finding is supported by the quantitative results. When nonlawyers and lawyers appear at hearings, they disclose evidence, present documentary evidence, present testimony at similar rates, and have similar case outcomes.
However, we also find an important limitation on nonlawyer expertise and effectiveness: they are not equipped to challenge or disagree with judges on substantive or procedural legal issues. Where lawyers seem to take for granted that they will disagree with judges from time to time, and that they will voice those disagreements in the context of advocating for their clients, nonlawyers' relationship to judges is clearly deferential. Nonlawyers' understanding of law and procedure is based on what they have learned from judges—most do no legal research and do not read procedural rules, statutes, or cases.
Only one nonlawyer reported seeking out and reading case law. This nonlawyer is unique among those interviewed, and we believe she is unique among nonlawyers at OAH more generally. Her statements demonstrate that she wants to understand unemployment law, but recognizes that such knowledge is often transmitted through judges. Discussing how nonlawyers stay informed about changes in the law, she said:
Nothing formal. We are not attorneys, so we're not always attuned. [I] read precedent but have a hard time finding [opinions]. [I] can't find them even on the [court] website. Judges can give me copies of decisions. They might pull me aside and say “did you hear the latest nonsense?” [In my firm, we] compare notes and keep an electronic folder of decisions as they come in … It's a challenge for us and everyone in the industry. [There is a] learning curve. We're not lawyers. I don't want to be one. I don't want to act like one.
This nonlawyer also notes: “Independent hearing reps tend to have little if any knowledge of [Court of Appeals] cases that come out.” We see her desire to learn the law but also her limitations in achieving that goal. Though she “reads precedent” when she can find it, she noted challenges in the very basic (for lawyers) process of locating appellate decisions.Footnote 38 Her statements also highlight judges' roles in transmitting information about the law, as well as the nature of her relationship to the judges.
In contrast, conversations with lawyers suggest their formal legal training, knowledge of law and procedure, and varied litigation experience (key ways in which they differ from nonlawyers) translate to a willingness to challenge judges on contested issues when necessary. The lawyers in the study expressed respect for the OAH judges, but did not hesitate to criticize them. Lawyers reported accounting for individual judges' personalities and preferences in case preparation and practice, indicating relational expertise. However, unlike nonlawyers, they did not report deferential behavior or describe judges as a source of their expertise. Lawyers reported challenging judges on points of law and procedure when necessary. As one said, “there are a number of different judges who see these cases differently. Some are royal pains, and some are laid back.” Another lawyer described how she routinely pushes judges on a particular contested point of law, stating that she will “pound the table” in closing argument as she makes the point, arguing for the judge to accept her view.
Lawyers view judges as an audience to be persuaded while nonlawyers explicitly see judges as teachers and primary sources of knowledge about court procedure and substantive law. As a result, nonlawyers seem to behave more like stewards than advocates. They understand the key steps of putting on a case (going through the process of introducing documents, presenting testimony, and putting facts on the record), but showed no inclination or ability to argue points of law or procedure with the judges, which is entirely consistent with how they were trained. The nonlawyers did not report reading the procedural rules that govern practice at OAH, let alone substantive law, but they do possess a clear understanding of what each judge expects in a hearing. In their work, the nonlawyers color within the lines drawn by the judges. Most nonlawyers possess little to no independent knowledge of law or procedure; thus it is hard to imagine how they could draw on law or procedure, particularly little-used or contested rules, to force the hand of the very judges who trained them.
We hasten to add that nonlawyers may be effective in many cases; indeed, the quantitative results on case outcomes suggest this is true. In civil litigation, raising procedural objections or arguing points of law is not necessary in every single case. Sometimes, an individual judge's preferences line up with a client's case and some cases do not involve contested legal or procedural issues. However, empirical evidence also demonstrates that judges often fail to follow or enforce procedural rules, and that courts systematically treat some cases in a perfunctory way (Sandefur Reference Sandefur2015). We also know that judges in general and judges at OAH specifically do not always interpret or apply the law correctly.Footnote 39 Finally, the law itself is sometimes unclear or does not contemplate a situation presented in a given case.
With these considerations in mind, if, as we find, judges play a primary role in training nonlawyers and shaping their expertise, can those same nonlawyers press judges to follow certain procedural rules (perhaps those a judge may prefer to ignore) or alter their interpretation of substantive law? Would they even be able to identify when such advocacy is necessary, let alone implement it? Our qualitative findings suggest they cannot. Such moves are simply inconsistent with how they were trained.
These findings add a layer of complexity to existing theories of legal expertise, suggesting the need for more research into how nonlawyers develop expertise in different civil justice contexts, and how this, in turn, shapes the exercise of expertise. While the nonlawyers in this study help their clients through knowledge of common court procedures and basic substantive law, their expertise and effectiveness has limitations. The extent to which nonlawyer expertise is shaped by judges and through “trial and error” has implications for nonlawyer advocacy programs and for future research on legal representation.
VI. Implications
The findings point to the need for additional evidence about and analysis of nonlawyer legal practice, not only from the perspective of individual clients but, critically, how nonlawyer representation of individual clients fits into the broader justice system. Here, we note three implications of this research, including a tension between individual client needs and broader justice interests, questions about the desired role of nonlawyer advocates in serving individual clients, and new information about case selection processes.
First, the data reveal a tension and potential downside of nonlawyer practice that has not been previously identified in the access to justice literature, something we identify as a law reform problem (Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016a). Any representative has the potential to serve the interests of an individual client in a case, and also broader systemic justice interests. For the nonlawyers in this study, their expertise development process makes them deferential to judges (relational expertise), and their understanding of law and procedure has been shaped by their interactions with judges (substantive expertise). Thus, they are not well-equipped to exercise the necessary expertise to engage in law reform activity that serves systemic interests, which, by definition, typically involves challenging judges, whether through arguments at trial, post-trial motions, or appeals. Elsewhere, we call such activity “system-focused challenges” (Shanahan, Carpenter, and Mark, Reference Shanahan, Carpenter and Mark2016a). If nonlawyers operating in a given civil justice context are able to, as we suggest, provide basic legal assistance, but play a limited role, or no role, in advancing legal change through system-focused challenges, what does this mean for our system of justice? Nonlawyer practice concentrated in a particular area of law may be able to fill a measure of unmet legal need, but would such practice ultimately result in a lack of systemic advocacy to the detriment of the very people the practice was designed to help? Taking a more hopeful view, is it possible to train nonlawyers to engage in system-focused challenges, or work with lawyers to do so? (Shanahan, Carpenter, and Mark, Reference Shanahan, Carpenter and Mark2016a).
In the pro bono context, Cummings and Sandefur (Reference Cummings and Sandefur2013) have raised a similar concern. They argue that the quality and effect of pro bono work should be evaluated on an individual as well as systemic level, and that scholars must identify whether and when pro bono lawyers “miss opportunities to turn individual cases into law reform efforts” so that the work may be given to, or coordinated with, lawyers who specialize in systemic advocacy. Similarly, nonlawyer practice should be evaluated not only on the basis of case outcomes, but also based on how it contributes to or limits legal change on a broader scale, and questions about law reform should inform the expansion of nonlawyer practice.
A second issue, closely related to the law reform challenge, is what role nonlawyers should play when they provide individual client services. Is it sufficient for nonlawyers to serve as stewards, helping clients navigate the basic steps of legal processes, or should they also be able to engage in legal advocacy for individual clients that challenges the judge in that proceeding to shape the law to fit the client's needs? We have termed this activity “case-focused challenges” (Shanahan, Carpenter, and Mark Reference Shanahan, Carpenter and Mark2016a). The nonlawyers in this study, although capable of helping their clients present a case and introduce evidence, are less skilled in making independent judgments and arguments about law and procedure to advance their clients' interests through case-focused challenges. For example, a case-focused challenge might involve making an argument that the facts of a client's case can be distinguished from facially similar facts in unfavorable controlling case law. Such advocacy based on substantive law is not an aspect of the expertise possessed by nonlawyers in this study. Despite this, is it possible to create a system where nonlawyers do have such training, or call on lawyers for assistance when they are out of their depth (Shanahan, Carpenter, and Mark, Reference Shanahan, Carpenter and Mark2016a)?
Third, and finally, the findings in this study complicate and challenge common assumptions about case selection processes in access to justice research. The idea that lawyers, as a general matter, select cases and clients primarily based on “merit” is pervasive in legal scholarship and has particular salience for empirical research on legal representation. Researchers, when discussing or accounting for selection effects, often assume that likelihood of winning is a driving factor in lawyers' case selection (Genn Reference Genn1993; Engler Reference Engler2009; Greiner and Pattanayak Reference Greiner and Pattanayak2012; Miller, Keith, and Holmes Reference Miller, Keith and Holmes2015). While this assumption is certainly reasonable and defensible, we lack empirical evidence about case selection processes across different types of lawyers and in the many different contexts where they practice.
The literature is also rife with other, often contradictory, theories about selection processes in civil justice. The following is a fairly representative sample: people seek lawyers when they think cases are complex (Sandefur Reference Sandefur2015) or when they think cases are sufficiently important (Rhode Reference Rhode2009). Nonprofit lawyers select cases based on perceived benefit to client (Seron et al. Reference Seron, Carroll, Gregg and Jean2001). And self-help programs select cases with merit and reject those without (Abel Reference Abel2010).
At least two pieces of empirical evidence, in addition to this study, present a mixed picture of lawyers' case selection processes. One study finds that lawyers take on more difficult cases, as opposed to the easiest cases (Monsma and Lempert Reference Monsma and Lempert1992), while another study finds that lawyers are overconfident in their litigation-outcome predictions (Goodman-Delahunty et al. Reference Goodman‐Delahunty, Par A, Maria and Elizabeth F2010). In our data, ideas about the “merits” of a case clearly play a role in some selection choices, but not all. Lawyers in our sample, including worker lawyers, choose cases for a variety of reasons: demographics; clients' perceived “need” of assistance (is the client disabled or low-literacy?); complexity of the case (how difficult would it be for a layperson to advocate for themselves); ripeness for law reform; personality (one lawyer stated she represents only clients she “likes” and finds deserving); and, finally, some lawyers essentially do not screen at all. Clearly, there are a range of views on how lawyers choose cases and there is no theoretical or empirical consensus on this issue.
On the HR firm and nonlawyer side, this study shows that nonlawyers themselves do not screen based on merit, they merely take cases when their calendars allow it. On the HR firm side, ideas about merit clearly play some role in the decision to attend a hearing; however, it is hard to disentangle HR firms' ideas about merit from the strong financial incentive not to attend hearings. Any finding that HR firms screen based on perceived merit must be viewed through the lens of nonlawyers' legal expertise. Our analysis suggests assessments of merit are questionable at best when done by lawyers, and that nonlawyer assessments of merit, given limited expertise in substantive law and procedure, are generally not meaningful. Across legal practice, the “merits” of a given client's case may be moot as it relates to that client's likelihood of prevailing. Default by defendants, plaintiffs' failure to prosecute, pressure to settle, and cases won based on evidentiary, procedural, or preliminary rulings are a common part of everyday legal practice.
To advance understanding of legal representation and its role in access to justice, scholars need to learn more about how representatives (lawyers and nonlawyers) select cases and why, not only to add context to observational research, but also to understand selection as a substantive function of legal representation and the civil justice system. The common assumption that lawyers select cases mainly based on merit should be questioned, if not rejected, in light of the complexity of the actual selection processes uncovered here. There may be much more randomness to case selection processes than has been acknowledged in the literature.
APPENDIX Procedural Behaviors—Correlation Table
| rep appears | party appear | disclose | intro/admit | testify | |
|---|---|---|---|---|---|
| rep appears | 1.0000 | ||||
| intro/admit docs | 0.5600* | 1.0000 | |||
| 0.0000 | |||||
| disclose docs | 0.4819* | 1.0000 | 0.6550* | ||
| 0.0000 | 0.0000 | ||||
| party appear | 0.5899* | 1.0000 | 0.5809* | 0.7704* | |
| 0.0000 | 0.0000 | 0.0000 | |||
| testify | 0.5477* | 0.8918* | 0.5172* | 0.7471* | 1.0000 |
| 0.0000 | 0.0000 | 0.0000 | 0.0000 |
