Automated, data‐driven decision making is increasingly common in a variety of application domains. In educational software, for example, machine learning has been applied to tasks like selecting the next exercise for students to complete. Machine learning methods, however, are not always equally effective for all groups of students. Current approaches to designing fair algorithms tend to focus on statistical measures concerning a small subset of legally protected categories like race or gender. Focusing solely on legally protected categories, however, can limit our understanding of bias and unfairness by ignoring the complexities of identity. We propose an alternative approach to categorization, grounded in sociological techniques of measuring identity. By soliciting survey data and interviews from the population being studied, we can build context‐specific categories from the bottom up. The emergent categories can then be combined with extant algorithmic fairness strategies to discover which identity groups are not well‐served, and thus where algorithms should be improved or avoided altogether. We focus on educational applications but present arguments that this approach should be adopted more broadly for issues of algorithmic fairness across a variety of applications.
This content will become publicly available on May 10, 2024
A Psychometric Framework for Evaluating Fairness in Algorithmic Decision Making: Differential Algorithmic Functioning
As algorithmic decision making is increasingly deployed in every walk of life, many researchers have raised concerns about fairness-related bias from such algorithms. But there is little research on harnessing psychometric methods to uncover potential discriminatory bias inside decision-making algorithms. The main goal of this article is to propose a new framework for algorithmic fairness based on differential item functioning (DIF), which has been commonly used to measure item fairness in psychometrics. Our fairness notion, which we call differential algorithmic functioning (DAF), is defined based on three pieces of information: a decision variable, a “fair” variable, and a protected variable such as race or gender. Under the DAF framework, an algorithm can exhibit uniform DAF, nonuniform DAF, or neither (i.e., non-DAF). For detecting DAF, we provide modifications of well-established DIF methods: Mantel–Haenszel test, logistic regression, and residual-based DIF. We demonstrate our framework through a real dataset concerning decision-making algorithms for grade retention in K–12 education in the United States.
more » « less- Award ID(s):
- 2225321
- NSF-PAR ID:
- 10412872
- Publisher / Repository:
- DOI PREFIX: 10.3102
- Date Published:
- Journal Name:
- Journal of Educational and Behavioral Statistics
- Volume:
- 49
- Issue:
- 2
- ISSN:
- 1076-9986
- Format(s):
- Medium: X Size: p. 151-172
- Size(s):
- ["p. 151-172"]
- Sponsoring Org:
- National Science Foundation
More Like this
-
more » « less
Abstract -
The Standards for educational and psychological assessment were developed by the American Educational Research Association, American Psychological Association, and National Council on Measurement in Education (AERA et al., 2014). The Standards specify assessment developers establish five types of validity evidence: test content, response processes, internal structure, relationship to other variables, and consequential/bias. Relevant to this proposal is consequential validity evidence that identifies the potential negative impact of testing or bias. Standard 3.1 of The Standards (2014) on fairness in testing states that “those responsible for test development, revision, and administration should design all steps of the testing process to promote valid score interpretations for intended score uses for the widest possible range of individuals and relevant sub-groups in the intended populations” (p. 63). Three types of bias include construct, method, and item bias (Boer et al., 2018). Testing for differential item functioning (DIF) is a standard analysis adopted to detect item bias against a subgroup (Boer et al., 2018). Example subgroups include gender, race/ethnic group, socioeconomic status, native language, or disability. DIF is when “equally able test takers differ in their probabilities answering a test item correctly as a function of group membership” (AERA et al., 2005, p. 51). DIF indicates systematic error as compared to real mean group differences (Camilli & Shepard, 1994). Items exhibiting significant DIF are removed or reviewed for sources leading to bias to determine modifications to retain and further test an item. The Delphi technique is an emergent systematic research method whereby expert panel members review item content through an iterative process (Yildirim & Büyüköztürk, 2018). Experts independently evaluate each item for potential sources leading to DIF, researchers group their responses, and experts then independently complete a survey to rate their level of agreement with the anonymously grouped responses. This process continues until saturation and consensus are reached among experts as established through some criterion (e.g., median agreement rating, item quartile range, and percent agreement). The technique allows researchers to “identify, learn, and share the ideas of experts by searching for agreement among experts” (Yildirim & Büyüköztürk, 2018, p. 451). Research has illustrated this technique applied after DIF is detected, but not before administering items in the field. The current research is a methodological illustration of the Delphi technique applied in the item construction phase of assessment development as part of a five-year study to develop and test new problem-solving measures (PSM; Bostic et al., 2015, 2017) for U.S.A. grades 6-8 in a computer adaptive testing environment. As part of an iterative design-science-based methodology (Middleton et al., 2008), we illustrate the integration of the Delphi technique into the item writing process. Results from two three-person panels each reviewing a set of 45 PSM items are utilized to illustrate the technique. Advantages and limitations identified through a survey by participating experts and researchers are outlined to advance the method.more » « less
-
Algorithmic decision-making systems are increasingly used throughout the public and private sectors to make important decisions or assist humans in making these decisions with real social consequences. While there has been substantial research in recent years to build fair decision-making algorithms, there has been less research seeking to understand the factors that affect people's perceptions of fairness in these systems, which we argue is also important for their broader acceptance. In this research, we conduct an online experiment to better understand perceptions of fairness, focusing on three sets of factors: algorithm outcomes, algorithm development and deployment procedures, and individual differences. We find that people rate the algorithm as more fair when the algorithm predicts in their favor, even surpassing the negative effects of describing algorithms that are very biased against particular demographic groups. We find that this effect is moderated by several variables, including participants' education level, gender, and several aspects of the development procedure. Our findings suggest that systems that evaluate algorithmic fairness through users' feedback must consider the possibility of "outcome favorability" bias.more » « less
-
When measuring academic skills among students whose primary language is not English, standardized assessments are often provided in languages other than English. The degree to which alternate-language test translations yield unbiased, equitable assessment must be evaluated; however, traditional methods of investigating measurement equivalence are susceptible to confounding group differences. The primary purposes of this study were to investigate differential item functioning (DIF) and item bias across Spanish and English forms of an assessment of early mathematics skills. Secondary purposes were to investigate the presence of selection bias and demonstrate a novel approach for investigating DIF that uses a regression discontinuity design framework to control for selection bias. Data were drawn from 1,750 Spanish-speaking Kindergarteners participating in the Early Childhood Longitudinal Study, Kindergarten Class of 1998–1999, who were administered either the Spanish or English version of the mathematics assessment based on their performance on an English language screening measure. Evidence of selection bias—differences between groups in SES, age, approaches to learning, self-control, social interaction, country of birth, childcare, household composition and number in the home, books in the home, and parent involvement—highlighted limitations of a traditional approach for investigating DIF that only controlled for ability. When controlling for selection bias, only 11% of items displayed DIF, and subsequent examination of item content did not suggest item bias. Results provide evidence that the Spanish translation of the ECLS-K mathematics assessment is an equitable and unbiased assessment accommodation for young dual language learners.more » « less
-
Given significant concerns about fairness and bias in the use of artificial intelligence (AI) and machine learning (ML) for psychological assessment, we provide a conceptual framework for investigating and mitigating machine-learning measurement bias (MLMB) from a psychometric perspective. MLMB is defined as differential functioning of the trained ML model between subgroups. MLMB manifests empirically when a trained ML model produces different predicted score levels for different subgroups (e.g., race, gender) despite them having the same ground-truth levels for the underlying construct of interest (e.g., personality) and/or when the model yields differential predictive accuracies across the subgroups. Because the development of ML models involves both data and algorithms, both biased data and algorithm-training bias are potential sources of MLMB. Data bias can occur in the form of nonequivalence between subgroups in the ground truth, platform-based construct, behavioral expression, and/or feature computing. Algorithm-training bias can occur when algorithms are developed with nonequivalence in the relation between extracted features and ground truth (i.e., algorithm features are differentially used, weighted, or transformed between subgroups). We explain how these potential sources of bias may manifest during ML model development and share initial ideas for mitigating them, including recognizing that new statistical and algorithmic procedures need to be developed. We also discuss how this framework clarifies MLMB but does not reduce the complexity of the issue.more » « less
