Predictive Bias in Graduate Admissions Tests

Swinton, Spencer S.

doi:10.1002/j.2333-8504.1981.tb01259.x

Cited by 3 publications

(3 citation statements)

References 46 publications

(40 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our analysis shows that prediction bias does exist in the sense that a single equation yields predictions for minority students with low predictor scores which, on average, tend to be somewhat higher than the criterion scores they actually achieve. This finding is consistent with much of the previous research in this area (Rolph, et al, 1978;Swinton, 1980). A more detailed discussion of the evidence for -2prediction bias in the GMAT program will be presented in Section 3.…”

Section: Introductionsupporting

confidence: 88%

“…The new system takes a student's race into account when predicting FYA. Some of the questions surrounding the use of demographic characteristics in prediction equations have been cogently summarized and discussed in a companion report by Swinton (1980). Our own view is that if such characteristics are considered at other stages in the admissions process, they should be permitted to playa role in producing better estimates of the candidate's FYA.…”

Section: Discussionmentioning

confidence: 99%

“…This notion of "fairness to the group" is only one way of posing the problem of prediction bias. Swinton (1980) provides an extensive discussion of other possible formulations as well as a lucid survey of the literature on the subject. Except for some simple illustrations in Section 2, our aim is neither to recapitulate Swinton's work nor to contribute to the foundational questions in the area.…”

Section: Overview ----mentioning

confidence: 99%

See 2 more Smart Citations

The Graduate Management Admission Test Prediction Bias Study

Braun¹,

Jones²,

Rubin³

et al. 1981

ETS Research Report Series

View full text Add to dashboard Cite

This document presents the results of an investigation into the prediction of first‐year average in graduate management school based on undergraduate grade point average and GMAT scores. It is found that the use of a single prediction equation for both black students and white students tends to produce over predictions of the former's graduate school performance. Substantial effort is devoted to developing a prediction system which takes into account ethnic status. This approach is made practicable by employing the empirical Bayes estimation paradigm, and the new predictions represent a substantial improvement over the ones based on a single equation for both races.

show abstract

Section: Introductionsupporting

confidence: 88%

Section: Discussionmentioning

confidence: 99%

Section: Overview ----mentioning

confidence: 99%

See 1 more Smart Citation

The Graduate Management Admission Test Prediction Bias Study

Braun¹,

Jones²,

Rubin³

et al. 1981

ETS Research Report Series

View full text Add to dashboard Cite

show abstract

Charities and politics ? The perennial problem

Lee

1984

Liverpool Law Rev

View full text Add to dashboard Cite

Predictive Biases in Natural Language Processing Models: A Conceptual Framework and Overview

Shah

Schwartz

Hovy

2020

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics

132

View full text Add to dashboard Cite

An increasing number of natural language processing papers address the effect of bias on predictions, introducing mitigation techniques at different parts of the standard NLP pipeline (data and models). However, these works have been conducted individually, without a unifying framework to organize efforts within the field. This situation leads to repetitive approaches, and focuses overly on bias symptoms/effects, rather than on their origins, which could limit the development of effective countermeasures. In this paper, we propose a unifying predictive bias framework for NLP. We summarize the NLP literature and suggest general mathematical definitions of predictive bias. We differentiate two consequences of bias: outcome disparities and error disparities, as well as four potential origins of biases: label bias, selection bias, model overamplification, and semantic bias. Our framework serves as an overview of predictive bias in NLP, integrating existing work into a single structure, and providing a conceptual baseline for improved frameworks.

show abstract

Predictive Bias in Graduate Admissions Tests

Cited by 3 publications

References 46 publications

The Graduate Management Admission Test Prediction Bias Study

The Graduate Management Admission Test Prediction Bias Study

Charities and politics ? The perennial problem

Predictive Biases in Natural Language Processing Models: A Conceptual Framework and Overview

Contact Info

Product

Resources

About