Considering Academic Achievement in Socioeconomic Context: New Directions for College Access

• In Event: 35.080 - Validity and Fairness Issues in Assessing Higher Education Students

Abstract

The SAT and ACT have come under intense public scrutiny in recent years (e.g., Sacks, 1997; Zwick, 2004), dogged by fears that these college admissions tests measure “the size of students’ houses” (Kohn, 2001, p. B12) as much as their academic potential. In response to widespread concerns about test validity and fairness, researchers have investigated a variety of alternatives, ranging from test score adjustments based on socioeconomic factors (Carnevale, Rose, & Strohl, 2014; Young & Johnson, 2004) to “test-optional” college admissions policies (Belasco, Rosinger, & Hearn, 2014). The purpose of this paper is to introduce one university’s approach to evaluating test scores in light of applicants’ socioeconomic status (SES) and to review college outcomes for students admitted under these new metrics.

In response to a 2008 ballot initiative seeking to outlaw race-based affirmative action in Colorado, the University of Colorado at Boulder (CU) began developing the Overachievement Indices: race-blind, class-conscious admissions metrics designed to support college access for disadvantaged students (Gaertner & Hart, 2013). First implemented in 2011, the Overachievement Indices compare applicants’ SAT and ACT scores to those of students from similar socioeconomic backgrounds. Applicants who exceed SES-based expectations earn a significant boost in the admissions process. Although similar class-based measures have been proposed in the literature (Studley, 2003), little is known about college academic outcomes for the students singled out by these measures. Therefore, this paper is motivated by the following research question: How well can we expect the beneficiaries of class-conscious admissions to perform in college?

Under controlled experimental conditions (see Gaertner & Hart, 2013), CU identified a set of “class-based admits” – students who would not have been admitted without the Overachievement Indices. Class-based admits were assigned proxies in both current and historical CU data using Coarsened Exact Matching (CEM; Iacus, King, & Porro, 2011). While current CU data reveal college performance trends for students admitted since 2011, historical data are needed to forecast long-term outcomes such as four- and six-year graduation rates.

Table 1 presents college outcomes for class-based admits and all other CU students (“Baseline”) in the current and historical data. As Table 1 suggests, class-based admits can be expected to keep pace at CU. While college outcomes for current class-based admits lag slightly behind the baseline, the historical data suggest that in the long term, class-based admits’ grades and graduation rates may surpass those of typical undergraduates.

Although a variety of remedies have been proposed to reduce the influence of SES on admissions test scores (e.g., Carnevale & Rose, 2004), almost no research exists cataloging the effects of putting those remedies into practice. This paper attempts to address that gap. Colorado’s Overachievement Indices offer one example of admissions metrics that are correlated with college performance, but by construction unrelated to SES. Current trends and forecasts at CU suggest that considering academic credentials in socioeconomic context may reduce the influence of SES on college access without seriously diminishing academic outcomes on campus.

Author

• Matthew Newman Gaertner, WestEd