Admissions Tests Discriminate Against College Admission of Minority and Low-Income Students at Selective Colleges

Sep 14, 2026

Uploaded by Ivy

SAT and ACT admission test scores follow a bell-curve pattern, and statistical properties of a normal distribution can magnify inequities at high-score extremes. Small differences in average test performance across student populations translate into large differences in the share reaching top scores, thereby affecting admissions rates at the most selective colleges. The analysis explains how shifting and skewing the distribution increases disparities beyond the mean, and validates the mechanism using theoretical background and real data.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages13

Reading Time33 min

AccessFree to download

Theoretical Background

Bell Curve Assumptions for Combined SAT Scores

Shifting the Bell Curve and Disparate Impact

Actual vs. Theoretical Score Distributions

Comparing NPSAS Data to the Model

Student Aid Policy Analysis Papers Admissions Tests Discriminate Against College Admission of Minority and Low-Income Students at Selective Colleges [www.studentaidpolicy.com/sat-and-selectivity/](www.studentaidpolicy.com/sat-and-selectivity/) Mark Kantrowitz President Admissions Tests Discriminate Against College Admission of Minority and LowIncome Students at Selective Colleges By Mark Kantrowitz SAT and ACT admission test scores follow a normal distribution, also known as a Bell Curve. Artifacts of a normal distribution contribute to discrimination against minority and low-income students at the most selective colleges and universities. In particular, shifting the location of the bell curve due to changes in average combined SAT test scores magnifies differences at the extremes much more so than differences at the mean. Skew in the distribution can yield a similar effect. These problems are more likely to occur at two or more standard deviations beyond the mean, such as when the combined SAT test scores are 1400 or more and ACT test scores are 31 or more. This means that small differences in average SAT and ACT test scores among any two student populations will lead to big differences in the percentage of students with high test scores among those populations, yielding big differences in college admissions rates at the most selective colleges. A discussion of the theoretical background illustrates the nature of the problem, and is followed by real data that demonstrates that the actual magnitude of the problem is consistent with the theoretical analysis. This paper not only demonstrates that the SAT and ACT tests contribute to discrimination against the college admission of minority and low-income students at selective colleges, but also articulates the mechanism through which this discrimination occurs. Theoretical Background The SAT involves math and verbal tests each of which is scored on a scale of 200 to 800. The two test scores are added to yield a combined SAT test score on a scale of 400 to 1600. Chart 1 shows a theoretical normal distribution for combined SAT test scores, assuming a mean combined SAT test score of 1000 with a standard deviation of 200. Even though the x axis is in increments of 50, the underlying data is in increments of 10. Chart 1 | Bell Curve Distribution for Combined SAT Test Scores
Mean 1000, Standard Deviation 200 | | | | --- | --- | --- | | Percent of Student Population | 2.50%
2.00%
1.50%
1.00%
0.50%
0.00% | | | 400
450
500
550
600
650
700
750
800
850
900
950
1000
1050
1100
1150
1200
1250
1300
1350
1400
1450
1500
1550
1600
Combined SAT Test Scores | | | These assumptions are close to the actual mean combined SAT test scores and standard deviations in recent years, as shown in Table 1, which is based on Table 226.40 from the 2019 Digest of Education Statistics1 and the College Board’s SAT Suite of Assessments Annual Reports.2 The SAT was initially designed with these assumptions, and the College Board recentered the test with a mean of 1000 and a standard deviation of 200 in 1995, fitting it to a normal distribution.3 Table 1 | Mean Combined
Year SAT Test Score Standard Deviation | | | | | --- | --- | --- | --- | | 2020 | 1051 | | 211 | | 2019 | 1059 | | 210 | | 2018 | 1068 | | 204 | | 2017 | | 1060 | 195 | 1 [https://nces.ed.gov/programs/digest/d19/tables/dt19_226.40.asp](https://nces.ed.gov/programs/digest/d19/tables/dt19_226.40.asp) 2 The College Board’s mean scores for verbal and math were calculated separately and then added, as opposed to calculating an average for the total score, as used in the NPSAS:16 data, which may yield a different result. 3 [https://www.ets.org/Media/Research/pdf/RR-02-04-Dorans.pdf](https://www.ets.org/Media/Research/pdf/RR-02-04-Dorans.pdf) Chart 2 shows what happens when the location of the Bell Curve is shi

FAQs

What does the paper use to validate its theoretical explanation?

It compares the theoretical normal-distribution model to actual data using studies such as NPSAS:16, showing consistency between the predicted magnitude of effects and observed score distributions.

Sep 14, 2026

Uploaded by Ivy