College GPA in personnel selection: Implications for adverse impact

Adjei, Kwesi Owusu

Permalink

https://hdl.handle.net/2142/125769 Copy

Description

Title

College GPA in personnel selection: Implications for adverse impact

Author(s)

Adjei, Kwesi Owusu

Issue Date

2024-07-02

Director of Research (if dissertation) or Advisor (if thesis)

Newman, Daniel

Doctoral Committee Chair(s)

Newman, Daniel

Committee Member(s)

• Briley, Ava
• Zhang, Bo
• Alexander, Leo
• Zhang, Susu

Department of Study

Psychology

Discipline

Psychology

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Adverse Impact
• Race
• College GPA

Abstract

In personnel selection systems, two of the most common goals are to hire high-performing individuals and to ensure a diverse group of new hires. Some organizations have begun to show reticence toward using cognitive tests, because--although they exhibit high criterion validity (Sacket et al., 2022)--they also display large mean racial gaps (Roth et al., 2011). Ployhart and Holtz (2008) suggest the possibility of using GPA as a proxy for cognitive tests, theorizing it could lessen adverse impact potential. Relatedly, Roth and Bobko (2000) noted a sizeable mean race gap in college GPA and observed (cross-sectionally) that this gap grows over the years of a student’s undergraduate career. In Study 1, I conducted a meta-analysis to determine the average magnitude and variability of the racial gap in college GPA. Results (k = 39, N = 1,639,135) suggested a large mean gap (3/5 of an SD), that is fairly consistent across date of publication, published vs. unpublished sources, self-report vs. university records, and 1st year vs. cumulative GPA. Results also reveal that the classic Roth and Bobko (2000) findings are somewhat of an outlier among other primary studies. In Study 2, I revisit the conclusions of Roth and Bobko (2000), using a large multi-university dataset (N = 21,597) with longitudinal GPA measures taken on the same individuals across four years, examining how the GPA gap changes over time. Multi-group latent growth modeling results reveal intercept differences, but no slope differences, between Black and White college students’ GPAs—in contrast to the classic findings of Roth and Bobko (2000). In Study 3, I tested a theoretical model for why GPA is related to job performance, supporting cognitive ability and conscientiousness as the key mechanisms. Also in Study 3, I used the meta-analytic estimate for the race gap in GPA from Study 1 to directly evaluate the implications of using GPA as proxy for cognitive tests, by evaluating the job performance and adverse impact outcomes of hiring on college GPA, both alone and in combination with cognitive tests and/or conscientiousness. Results show that using college GPA as a proxy for cognitive tests yields lower criterion validity but higher diversity, and that college GPA holds little incremental value for job performance prediction or adverse impact reduction if used in conjunction with cognitive tests and Conscientiousness. Implications and limitations of hiring on college GPA for adverse impact reduction are discussed.  

Graduation Semester

2024-08

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/125769

Copyright and License Information

Copyright 2024 Kwesi Adjei

Owning Collections