Disparate Impact Analysis Research Articles

Mitigating Bias in AI-Driven Recruitment : The Role of Explainable Machine Learning (XAI)

This article explores the critical role of Explainable Artificial Intelligence (XAI) in mitigating bias within AI-driven recruitment processes. As AI becomes increasingly prevalent in hiring practices, concerns about algorithmic bias and fairness have emerged. The article discusses how XAI techniques, such as SHAP and LIME, can be used to detect and interpret potential biases in recruitment algorithms. It examines the implementation of XAI for feature importance analysis, algorithmic bias detection, and disparate impact analysis across different demographic groups. The article addresses the challenges of balancing model complexity with explainability and the limitations of XAI in identifying systemic biases. By implementing XAI strategies, organizations can enhance the fairness and transparency of their hiring practices, ultimately fostering more diverse and equitable workplaces.

Responsible AI in automated credit scoring systems

In the last few years, Artificial Intelligence (AI) has achieved a notable momentum that, may deliver the expectations over many application sectors across the field. For this to occur, expert systems and rule-based models need to overcome the limitation of fairness and interpretability. Paradigms underlying this problem fall within the so-called explainable AI (XAI) field. This report presents the work on German credit card dataset to overcome the challenges of fairness, bias and in return, deem the machine learning models giving a responsible expectation. This is defined as responsible AI in practice. Since the dataset we dealt with, is to classify credit score of a user as good or bad, using fair ML modelling approach, the key metric of interest is the F1-score to reduce share of misclassifications. It is observed that hyper parameter tuned XGBoost model (GC2) gives optimal performance in terms of both F1-score, accuracy and fairness for the case of both gender and age as protected variable through Disparate Impact Remover, a pre-processing bias mitigation technique. The same is deployed using both Heroku through Flask API (for age). The Disparate Impact Analysis (DIA) using H2O.AI helped to identify optimum threshold levels at which the fairness metrics are observed at legally acceptable/permissible levels for both age and gender. Overall, fairness, bias responsibility and explainability have been established for the dataset considered.

Disparate Impact Analysis of the Critical Hire-Personality Assessment

Pre-employment personality testing has become widely used across a variety of work settings. A relatively new area where pre-employment personality testing is being implemented is in the field of community-based corrections (CBCs). One instrument being used in CBCs for this purpose is the Critical Hire-Personality Assessment (CH-PA). Federal laws and hiring guidelines have been enacted that impact the use of pre-employment personality assessments such as the CH-PA. The Age Discrimination in Employment Act and Title VII of the Civil Rights Act of 1964, for example, prohibit employment discrimination based on gender, age, and racial/ethnic status. The present study examines if the CH-PA complies with federal laws and guidelines regarding employee selections by examining the degree to which the CH-PA scale and subscale scores differ based on CBC officer gender, age, and racial/ethnic status. Results revealed that the CH-PA scales and subscales do not differ in a manner that would significantly contribute to a disparate or adverse impact based on CBC officer applicant gender, age, and racial/ethnic status. Implications of these findings for CBC agencies are discussed.

A Half Century Post Title VII: Still Seeking Pathways for Women to Organizational Leadership

Perhaps the most important of the many 50th anniversaries marked in 2014 is the passage of Title VII of the Civil Rights Act of 1964 (Title VII). Although Title VII and accompanying legislation and judicial rulings have made significant headway in improving the work environment for women, pathways for women to positions of leadership in organizations are still generally elusive. Our studies suggest additional challenges for women with dependents. In light of the overwhelming evidence that leadership opportunities are effectively denied to a large number of female citizens, it is time to reinterpret Title VII in a way to help remedy this type of discrimination. We thus advocate three proposals for reform. First, recognizing that quotas would be difficult to implement in the United States, we argue that in cases alleging gender discrimination, courts consider the paucity of women in leadership positions as a rebuttable presumption of discrimination – as a logical extension to the disparate impact analysis firmly established in precedent. Second, we implore the SEC to define diversity, in reports already required of public companies, to include gender diversity. Finally, we urge firms to improve programs for mentoring and networking to provide true opportunities across cultural backgrounds to break the logjam in the pathway for women seeking upward mobility.

Help Me, I'm Drowning!

This Note will demonstrate that the FHA and ECOA prohibit implementation of lenders’ threats to limit or refuse credit availability because the result would have a disproportionate effect on minorities. This Note will examine the legality of the lenders’ threats to withhold credit and, in doing so, presumes that courts will find this novel use of eminent domain constitutional. First, Part II will discuss the current situation. Section IIA will explain eminent domain and explore precisely how cities hope to use it in the mortgage note context. Section IIB will describe the growth of this plan and the cities that have considered using it. Section IIC will elaborate on the threats that the lenders are making toward the cities that consider using eminent domain. Second, Part III will discuss the various laws that regulate consumer transactions and how they have restricted lenders in the past. Section IIIA will give a brief history of redlining and its consequences. Section IIIB will discuss the circumstances surrounding the enactment of the FHA and ECOA and exactly what actions these statutes prohibit. Section IIIC will discuss the three causes of action that arise under the FHA and the ECOA. Section IIID will further explore disparate impact and how courts have interpreted 24 C.F.R. Section 100.500, which was enacted on March 18, 2013.18 Lastly, Part IV will apply disparate impact to the lenders’ threats and work through the disparate impact analysis in detail to analyze the likelihood that the FHA and the ECOA will be successful in restricting the lenders’ threats.

In Defense of Disparate Impact: Urban Redevelopment and the Supreme Court's Recent Interest in the Fair Housing Act

Twice in the past three years, the Supreme Court has granted certiorari in Fair Housing cases, and, each time, under pressure from civil rights leaders who feared that the Supreme Court might narrow current Fair Housing Act jurisprudence, the cases settled just weeks before oral argument. Settlements after the Supreme Court grants certiorari are extremely rare, and, in these cases, the settlements reflect a substantial fear among civil rights advocates that the Supreme Court’s recent decisions in cases such as Shelby County v. Holder and Fisher v. University of Texas are working to dismantle many of the protections of the Civil Rights legislation of the 1960s. The sole issue in both of the recently settled Fair Housing Act cases was whether disparate impact analysis – a type of analysis that some on the Supreme Court may view as requiring racial preferences – is valid under the Fair Housing Act. This article argues that in order to have a chance at achieving the goal of its sponsors – “to replace the ghettos [with] truly integrated and balanced living patterns,” – the Fair Housing Act cannot just take aim at the aberrant individual who intentionally denies a person housing because of his or her race. Instead, the Fair Housing Act must recognize claims based on disparate impact analysis alone. This article argues that disparate impact analysis is especially needed to address urban redevelopment decisions because such decisions are often made through a multi-party protracted process, in which a discriminatory intent may be impossible to discern or entirely absent. Additionally, it is the outcome of large-scale urban redevelopment projects that will truly shape racial housing patterns in the twenty-first century.

The Legal and the Local: Using Disparate Impact Analysis to Understand the Consequences of Writing Assessment

In this article, we investigate disparate impact analysis as a validation tool for understanding the local effects of writing assessment on diverse groups of students. Using a case study data set from a university that we call Brick City University, we explain how Brick City’s writing program undertook a self-study of its placement exam using the disparate impact process followed by the Office for Civil Rights of the US Department of Education. This three-step process includes analyzing placement rates through (1) a threshold statistical analysis, (2) a contextualized inquiry to determine whether the placement exam meets an important educational objective, and (3) a consideration of less discriminatory assessment alternatives. By employing such a process, Brick City re-conceptualized the role of placement testing and basic writing at the university in a way that was less discriminatory for Brick City’s diverse student population.

An Analysis of the U.S. Supreme Court’s Decision in Ricci v. DeStefano

Almost 6 months after winning their U.S. Supreme Court case, a group of New Haven, Connecticut, firefighters celebrated their victory in grand style. The decision in Ricci v. DeStefano proved that the City of New Haven erred when it denied promotions to White firefighters who fared better on promotional examinations than did minority applicants. This article (a) examines the thrust of the Ricci decision, which involved two competing facets of the same equal employment opportunity (EEO) statute; (b) discusses the mistakes inherent in the testing procedures of the City of New Haven, Connecticut; (c) addresses the involvement of Justice Sonya Sotomayor in the decision; and (d) examines the implication of the decision for public-sector employers using testing procedures of the nature found in Ricci v. DeStefano. The author concludes that the decision underscores the need for use of alternative testing procedures and suggests that Ricci v. DeStefano is but a precursor to further litigation designed to render use of disparate-impact analysis unconstitutional.

A Social Movement History of Title VII Disparate Impact Analysis

The U.S. Supreme Court’s recent opinion in Ricci v. DeStefano suggests trouble ahead for disparate impact analysis under Title VII of the Civil Rights Acts of 1964 and 1991. Commentators, too, have begun to question the policy bases for this doctrine. Part of the current tenuousness surrounding disparate impact analysis, which the Court first approved in its 1971 opinion in Griggs v. Duke Power Company, stems from assumptions that the EEOC pursued this theory as a last-minute, ill-conceived afterthought that was not in keeping with Congress’s intent when it passed Title VII in 1964. In this Article I use the methodology of social movements theory to cast a different light on the history of disparate impact analysis. Disparate impact analysis asks employers to evaluate whether their employment practices pose barriers to the employment opportunities of traditionally excluded outsiders and, if so, to replace such practices with alternative effective practices with less exclusionary effect. Disparate impact analysis thus augments federal employment anti-discrimination law by introducing a structural approach alongside an intent-based standard that focuses exclusively on detecting invidious motives buried in individual decision-makers’ hearts. In this Article I trace concerns about tackling structural race discrimination in employment to the early history of civil rights activism. I locate the origins of disparate impact analysis in civil rights activists’ long-held understandings about the structural nature of racial employment subordination. Throughout a long historical period, employment subordination based on race was pervasive and unremitting, enforced by a complex of institutions including law, tradition, and white violence insisting on a “whites only” rule for desirable jobs. These forces imposed a tight structural block on the employment advancement of persons of color, and civil rights activists therefore aimed their efforts at achieving broad-scale reform of these institutional barriers. Law also shaped the strategies of these early civil rights activists. In the late nineteenth and early twentieth centuries, the relevant legal question had nothing to do with the intent versus effects tests so heatedly debated today. Instead, two lines of the Court’s jurisprudence were important: its Post Reconstruction civil rights jurisprudence, which limited the reach of anti-discrimination law to state actors; and its Lochner Era employment law jurisprudence, which recognized a natural law “freedom of contract” right barring most forms of state regulation of the employment relationship. These two lines of caselaw defined the sphere of private employment as largely unreachable through federal or state law. Activists accordingly concentrated their efforts on inducing employers to voluntarily reform their workplace hiring and promotion practices. The end of the Lochner Era paved the way for New York State’s passage of the first employment anti-discrimination statute to reach the private sector. State agency officials charged with the statute’s enforcement, some recruited from the ranks of the activists responsible for the statute’s enactment, continued to experiment with methods of inducing employers – now with the background threat of legal action to encourage compliance – to make broad-scale institutional change by examining traditional employment practices to find ways to eliminate barriers to racial inclusiveness. By the time Congress began to debate Title VII’s passage, at least several prototypical disparate impact cases had emerged from state fair employment agencies. After the legislation’s enactment, the EEOC followed in the footsteps of state fair employment agencies by focusing its efforts on promoting broad-scale institutional reform. The EEOC encouraged employers in targeted industries to engage in self study to identify employment practices, such as testing and other selection devices, that could be reformed to increase employment opportunities for racial minorities. At the same time, the NAACP and others successfully litigated a number of disparate impact cases in the lower courts. At all these stages, civil rights activists and those charged with implementing statutes enacted as a result of these activists’ efforts developed and passed on ideas now embedded in the current statutory articulation of the disparate impact doctrine. Disparate impact analysis thus was not a last-minute afterthought, unwittingly endorsed by the U.S. Supreme Court, but the product of a sustained social movement effort over decades of theorizing and experimentation. In sum, my account adds important historical context to the current debate about the continued viability of Title VII disparate impact analysis and also illustrates the potential substantive payoff of a social movement focus on the development of public law.

Hiring Practices of African American Males in Academic Leadership Positions at American Colleges and Universities: An Employment Trends and Disparate Impact Analysis

This study examined the status of African American males in academic leadership positions at American colleges and universities in comparison with other males (e.g., Asian). Guided by disparate impact theory, descriptive trend analyses and impact ratios were computed using the 1993 and 1999 National Study of Postsecondary Faculty (NSOPF). These national-level data demonstrated that, although White males continue to hold the overwhelming majority of academic leadership positions, African American males have made modest advancements in upper level administrative positions at specific types of higher and postsecondary education institutions (e.g., 2-year and private institutions). More specifically, these results suggested that a disparate impact exists between the hiring of African American males and White males in academic leadership positions.

Hiring Practices of African American Males in Academic Leadership Positions at American Colleges and Universities: An Employment Trends and Disparate Impact Analysis

This study examined the status of African American males in academic leadership positions at American colleges and universities in comparison with other males (e.g., Asian). Guided by disparate impact theory, descriptive trend analyses and impact ratios were computed using the 1993 and 1999 National Study of Postsecondary Faculty (NSOPF). These national-level data demonstrated that, although White males continue to hold the overwhelming majority of academic leadership positions, African American males have made modest advancements in upper level administrative positions at specific types of higher and postsecondary education institutions (e.g., 2-year and private institutions). More specifically, these results suggested that a disparate impact exists between the hiring of African American males and White males in academic leadership positions.

Crime statistics, disparate impact analysis, and the economic disenfranchisement of minority ex-offenders

Crime statistics, disparate impact analysis, and the economic disenfranchisement of minority ex-offenders

Physical Ability Tests and Title VII

Suppose you supervise the police department of a large American city. In response to rising crime, the city council has authorized you to hire more officers. Within days your office is flooded with applications for just a handful of positions. How can you possibly choose from among them? One possibility is to screen those applicants initially to eliminate those who seem unlikely to become successful candidates. Imposing minimum education and literacy requirements might remove some applicants from consideration. Another option would be to screen based on physical ability. After all, many police officers must apprehend fleeing criminals, and skills such as strength and stamina could be measured by requiring applicants to drag or lift a certain amount of weight or complete a timed run. However, such a test might run afoul of employment discrimination law, specifically Title VII. Title VII protects minorities and women from employment practices discriminate against them. Since men tend to outperform women on most physical tests, more men than women will be hired as a result of using such tests. Consequently, tests based on physical strength can create an unlawful disparate impact. The Civil Rights Act of 19911 provides for a or defense to disparate impact claims under Title VII.2 Many employers rely on this defense when contesting their liability. The Act allows an employer to prevail if it proves that the challenged practice is job related for the position in question and consistent with business necessity.3 However, the exact contours of this defense are still unclear, and lower courts have used a number of different standards when evaluating the defense, thereby confusing plaintiffs and defendants alike. Part I of this Comment reviews the history of disparate impact analysis and the business necessity defense under Title VII and the Civil Rights Act of 1991. Part II examines one area where the business necessity defense crops up quite frequently, the use of physical ability

Insurance Redlining and Disparate Impact: The Application of a Federal Civil Rights Doctrine to Homeowners Insurance

ABSTRACT: Allegations of inner‐city insurance redlining are increasingly facilitated through the jurisprudence of “disparate impact,” a legal doctrine holding that a policy or practice based on race‐neutral criteria may nevertheless constitute illegal discrimination if it has a disproportionate adverse impact on racial minorities or women. Disparate impact analysis would require insurers to document a precise cause‐and‐effect relationship between a challenged underwriting variable and its associated risk. Moreover, they would be required to show that no “less discriminatory” risk‐assessment technique is available. If it is not possible—or too costly—to meet this burden, insurers will have no choice but to abandon the use of those risk‐selection practices and cost‐based pricing mechanisms that yield a disparate racial impact. This will result in higher premiums and less insurance availability for consumers. Furthermore, dubious charges of unfair discrimination will exacerbate racial tensions and divert attention from the social and economic pathologies of which insurance costs are merely symptomatic.