English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Colleges and universities use predictive analytics in a variety of ways toincrease student success rates. Despite the potential for predictive analytics,two major barriers exist to their adoption in higher education: (a) the lack ofdemocratization in deployment, and (b) the potential to exacerbateinequalities. Education researchers and policymakers encounter numerouschallenges in deploying predictive modeling in practice. These challengespresent in different steps of modeling including data preparation, modeldevelopment, and evaluation. Nevertheless, each of these steps can introduceadditional bias to the system if not appropriately performed. Most large-scaleand nationally representative education data sets suffer from a significantnumber of incomplete responses from the research participants. While manyeducation-related studies addressed the challenges of missing data, little isknown about the impact of handling missing values on the fairness of predictiveoutcomes in practice. In this paper, we set out to first assess the disparitiesin predictive modeling outcomes for college-student success, then investigatethe impact of imputation techniques on the model performance and fairness usinga commonly used set of metrics. We conduct a prospective evaluation to providea less biased estimation of future performance and fairness than an evaluationof historical data. Our comprehensive analysis of a real large-scale educationdataset reveals key insights on modeling disparities and how imputationtechniques impact the fairness of the student-success predictive outcome underdifferent testing scenarios. Our results indicate that imputation introducesbias if the testing set follows the historical distribution. However, if theinjustice in society is addressed and consequently the upcoming batch ofobservations is equalized, the model would be less biased.

Auditing the Imputation Effect on Fairness of Predictive Analytics in Higher Education