s
1
New Evidence on Recent
Changes in College Applications,
Admissions, and Enrollments
Focus on the Fall 2021 Admissions Cycle
Jessica Howell
Michael Hurwitz
Sam Imlay
Greg Perfetto
July 2022
s
2
Executive Summary
The Admissions Research Consortium (ARC) is a collaborative research initiative that
aims to help participating institutions gain insight into their own and their peers’
admissions processes and outcomes, as well as changes in student behavior influenced
by the covid-19 pandemic. Guided by a Core Advisory Committee (with representatives
from the Association for Institutional Research [AIR], American Association of Collegiate
Registrars and Admissions Officers [AACRAO], National Student Clearinghouse [NSC],
and senior admission and enrollment practitioners) and a Research Advisory Committee
composed of academic and institutional researchers, evidence and insights from ARC will
inform future practice and policy in the years following the pandemic. This Research Brief
presents initial evidence from ARC on changes in application, admission, and enrollment
trends in the fall 2021 college application cycle compared to prior years, and evidence on
applicants’ decisions about whether or not to submit their test scores for consideration in
the college admissions process.
Over 50 colleges, representing a range of selective public and private nonprofit four-year
institutions in the U.S., provided data on their applications, admissions, and enrollments
from fall 2018 to fall 2021. This information was merged with College Board assessment
data to enable research on how college-going trends and outcomes were affected by
pandemic-related disruptions. The data and analyses presented in this Research Brief are
meant to inform admissions practitioners of point-in-time trends across a subset of
institutions. The evidence does not necessarily generalize to all higher education
institutions, nor should the patterns documented in fall 2021 be viewed as definitively
stable in future years given the potential on-going and lasting effects of the pandemic on
both students and institutions. ARC will continue as a multi-year research initiative to
better understand longer-term trends and outcomes. Our research efforts will continue to
rely on data from ARC institutions, expand to examine data that more broadly represent
all higher education institutions, and incorporate findings from other researchers working
in this space.
Three themes emerged from the initial analyses of the fall 2021 college application cycle.
Theme 1: Between fall 2020 and fall 2021, the number of applications, offers of
admission, and students enrolled increased at institutions participating in ARC,
and at rates above and beyond prior years. Nearly all student subgroups
experienced increases in applications, offers of admission, and enrollment
between fall 2020 and fall 2021.
Applications to ARC institutions increased 17.8% between fall 2020 and fall 2021, with
application growth experienced by 96% of ARC institutions and across all types of ARC
s
3
institutions. While application growth outpaced growth in offers of admission and growth
in enrollment, offers of admission from ARC institutions increased by 6.2% between fall
2020 and fall 2021, with admissions growth experienced by 65% of ARC institutions and
across all types of ARC institutions except for More Selective Private ARC institutions.
Enrollment in ARC institutions increased by 7.9% between fall 2020 and fall 2021, with
increases in all types of institutions and enrollment growth experienced by 84% of ARC
institutions.
Applications to ARC institutions grew for all student subgroups between fall 2020 and fall
2021. Admissions offers from ARC institutions also grew for all student subgroups except
international students and students from the lowest two high school GPA quintiles.
Enrollment at ARC institutions grew between fall 2020 and fall 2021 for all student
subgroups except those from the lowest two high school GPA quintiles.
Theme 2: The composition of ARC applicants, admits, and enrollees, as measured
by the proportional representation of student subpopulations, changed very little
between fall 2018 and fall 2021, although there is variation across ARC institutions.
While applications to ARC institutions grew 17.8% last year, the share of under-
represented minority (URM) student applications to ARC institutions remained flat at 26%
over time due to similar growth in URM and non-URM student applications. The share of
URM students among admitted students to ARC institutions rose roughly 1 percentage
point to 26% in fall 2021 after being flat at 25% in prior years. URM student
representation among enrolled students at ARC institutions increased by 0.5 percentage
point each year since fall 2018, which suggests there was no overall break from prior
trends between fall 2020 and fall 2021. Considerable variation in minority student
representation exists across ARC institutions, with roughly half of institutions experiencing
increases and the other half experiencing decreases in URM student representation
between fall 2020 and fall 2021. Of the four ARC segments, More Selective Private ARC
institutions experienced the largest growth in the share of URM enrollees between fall
2020 and fall 2021.
Among the various proxies for lower-socioeconomic status (first generation, low-income,
and students from more disadvantaged neighborhoods), ARC institutions experienced
either no change or decreases over time in the representation of lower-socioeconomic
status students. As with URM student representation, there is considerable variation
across ARC institutions such that roughly half of institutions experienced increases and
half experienced decreases in lower-socioeconomic status student representation
between fall 2020 and fall 2021. Across all types of ARC institutions, More Selective
Private ARC institutions experienced the largest growth in the share of lower-
socioeconomic status enrollees between fall 2020 and fall 2021.
s
4
Theme 3: Approximately half of applications to ARC institutions in fall 2021
included SAT/ACT scores that students chose to disclose, another 30% of
applications withheld SAT scores, and roughly 20% had no recorded SAT score but
may have withheld an ACT score. Students’ SAT scores (relative to the college to
which they apply) are the strongest single predictor of their decision to disclose a
score in the application process.
Among the nearly 1.5 million applications to ARC institutions for fall 2021 entry, nearly
50% disclosed an SAT/ACT score, while nearly 30% withheld an SAT score and 20% had
no recorded SAT score. Test score disclosers had higher average test scores and higher
average high school grades than test score withholders.
Regression analyses reveal that test score, relative to test scores at the college to which
students apply, is the strongest determinant of a student’s decision to disclose a score,
where higher scoring students disclose scores at higher rates than lower scoring
students, on average. High school grades and student demographics add little additional
information about who discloses scores. Practically, these regression results imply that
students with very similar test scores and high school grades make very similar score
disclosure decisions when applying to ARC institutions regardless of other demographics
like race/ethnicity, first generation status, socioeconomic status. Previously published
differences in score disclosure patterns by race, parental education, and income are
attributable to differences in academic achievement among score disclosers and
withholders, for which those analyses were unable to control (Freeman et al., 2021).
s
5
Table of Contents
Executive Summary .................................................................................................2
The Admissions Research Consortium (ARC) .......................................................8
Guiding Principles for Interpreting ARC Data ............................................................ 8
ARC Data Sample and Definitions ............................................................................. 9
ARC Data Sample .................................................................................................................................. 9
ARC Data Definitions ............................................................................................................................ 10
ARC Fall 2021 Admissions Cohort: Aggregate Changes in College
Applications, Admissions Offers, and Enrollment .............................................. 12
Aggregate Enrollment Funnel Insights ..................................................................... 12
Aggregate Insights into Racial and Economic Diversity ........................................... 21
Aggregate Insights into Test Score Disclosure and Withholding ............................. 25
Conclusion .............................................................................................................. 32
Appendix ................................................................................................................. 33
Test Score Disclosure Regression Details ............................................................... 35
References .............................................................................................................. 36
About the College Board ........................................................................................ 36
s
6
List of Tables
Table 1: Percentage Changes in Applications Between Fall 2020 and Fall 2021, by
ARC Segment and Student Characteristics ............................................................. 16
Table 2: Percentage Changes in Admission Offers Between Fall 2020 and Fall 2021,
by ARC Segment and Student Characteristics ........................................................ 18
Table 3: Percentage Changes in Enrollment Between Fall 2020 and Fall 2021, by
ARC Segment and Student Characteristics ............................................................. 20
List of Figures
Figure 1: ARC Colleges’ Enrollment Funnel, Fall 2018-2021 .................................. 12
Figure 2: Enrollment Funnel for ARC Colleges and Segments, Fall 2018-2021 ...... 13
Figure 3: Percentage Change in Applications, Admissions, and Enrollment Between
Fall 2020 and Fall 2021, ARC Colleges and Segments ........................................... 14
Figure 4: Percentage Change in Applications to ARC Colleges Between Fall 2020
and Fall 2021, by Student Characteristics ............................................................... 15
Figure 5: Percentage Change in Admission Offers at ARC Colleges Between Fall
2020 and Fall 2021, by Student Characteristics ...................................................... 17
Figure 6: Percentage Change in Enrollment at ARC Colleges Between Fall 2020 and
Fall 2021, by Student Characteristics ...................................................................... 19
Figure 7: Racial/Ethnic Composition of Enrolled Students from Fall 2018 to Fall
2021, ARC Colleges ................................................................................................ 21
Figure 8: Racial/Ethnic Composition of Enrolled Students in Fall 2020 and Fall 2021,
ARC Colleges and ARC Segments ......................................................................... 22
Figure 9: Socioeconomic Composition of Enrolled Students from Fall 2018 to Fall
2021, ARC Colleges ................................................................................................ 23
Figure 10: Socioeconomic Composition of Enrolled Students in Fall 2020 and Fall
2021, ARC Colleges and ARC Segments................................................................ 24
Figure 11: Score Disclosure, Withholding, and Absence Among Fall 2021
Applicants, ARC Colleges and Segments................................................................ 26
Figure 12: Predictors of Test Score Disclosure for Fall 2021 ................................... 27
Figure 13: Probability of Test Score Disclosure Among ARC College Applications for
Fall 2021 .................................................................................................................. 28
Figure 14: Probability of Test Score Disclosure Among ARC College Applications for
Fall 2021, by HSGPA .............................................................................................. 29
s
7
Figure 15: Probability of Test Score Disclosure Among ARC College Applications for
Fall 2021, by Parental Education ............................................................................. 30
Figure 16: Probability of Test Score Disclosure Among ARC College Applications for
Fall 2021, by Race/Ethnicity .................................................................................... 30
s
8
The Admissions Research Consortium (ARC)
The Admissions Research Consortium (ARC) is a collaborative research initiative that aims
to help participating institutions gain insight into their own and their peers’ admissions
processes and outcomes, as well as changes in student behavior influenced by the covid-19
pandemic. Guided by a Core Advisory Committee (with representatives from the Association
for Institutional Research [AIR], American Association of Collegiate Registrars and
Admissions Officers [AACRAO], National Student Clearinghouse [NSC], and senior
admission and enrollment practitioners) and a Research Advisory Committee composed of
academic and institutional researchers, evidence and insights from ARC will inform future
practice and policy in the years following the pandemic. This Research Brief presents initial
evidence from ARC on changes in application, admission, and enrollment trends in the fall
2021 college application cycle compared to prior years.
Over 50 colleges, representing a range of selective public and private nonprofit four-year
institutions in the U.S., provided data on their applications, admissions, and enrollments
from fall 2018 to fall 2021. This information was merged with College Board assessment
data to enable research on how college-going trends and outcomes were affected by
pandemic-related disruptions. The data and analyses presented in this Research Brief are
meant to inform admissions practitioners of point-in-time trends across a subset of
institutions. The evidence does not necessarily generalize to all higher education
institutions, nor should the patterns documented in fall 2021 be viewed as definitively stable
in future years given the potential on-going and lasting effects of the pandemic on both
students and institutions. ARC will continue as a multi-year research initiative to better
understand longer-term trends and outcomes. Our research efforts will continue to rely on
data from ARC institutions, expand to examine data that more broadly represent all higher
education institutions, and incorporate findings from other researchers working in this space.
Guiding Principles for Interpreting ARC Data
There has never been a college application cycle like fall 2021. When examining changes
brought about by the pandemic, it is critical to recognize the many things simultaneously
affecting students and institutions: a global health crisis, a domestic economic crisis,
learning losses, mental health challenges, changes in opportunities to take standardized
assessments, and changes to college applications processes and practices including a
near-universal shift to test optional admissions policies that allowed students the opportunity
to choose whether to disclose or withhold their standardized test scores when applying. The
ARC Core Advisory Committee and Research Advisory Committee members have
cautioned about the importance of interpreting all data with care according to three
principles.
s
9
1. Avoid causal interpretations of descriptive, correlational data as well as
confirmation bias. With so many factors changing simultaneously, the analyses in this
Brief describe the combined effect of all factors on students and colleges and do not
provide causal evidence on the impact of any one factor on students or colleges. The
data in this Brief are presented in a straight-forward manner and without interpretation
that confirms any perspective or belief. The data are intended to fuel discussion and
further research, both quantitative and qualitative.
2. Be clear about what is not measurable or not visible in the data and where it is
impossible to ensure that analyses are comparing apples to apples. The pandemic
created disruptions to learning, mental health, physical health, opportunities to test, etc.
in ways that were not uniformly experienced – geographically, socioeconomically, or
demographically. Most of these factors are likely to influence choices and outcomes, yet
impossible to measure in existing data and thus may conflate results.
3. Recognize the dynamic nature of the moment and avoid the urge to craft a
definitive narrative based on one point in time or a single data point. Because the
impacts of the covid-19 pandemic are multi-dimensional and not uniformly experienced,
the data in this Brief must be viewed as part of a larger story that is still unfolding as we
continue to study (a) how the 2021 cohort progresses through college, (b) how future
cohorts of students navigate the college-going process, and (c) how college policies and
practices continue to change. The student and college behavior documented in this Brief
is still evolving in response to the pandemic, longstanding educational disparities, and
the interaction of those factors.
ARC Data Sample and Definitions
ARC Data Sample
In 2021, ARC institutions shared administrative data on applications, admissions, and
enrollments from fall 2018 to fall 2021. These data, which also include a robust set of
student demographic and academic variables, were merged to College Board assessment
data to enable insight into students who disclosed and withheld SAT test scores in the first
year of widespread test optional admissions policies brought on by the pandemic. This
Research Brief provides initial evidence on student and college choices based on the
combined dataset, which covers a sample of 51 four-year public and private nonprofit
institutions.
s
10
We categorize these 51 institutions into four ARC institutional segments defined by
institutions’ sector and selectivity:
1
• More Selective Private Colleges: 16 private institutions with admit rates below 25%
• Selective Private Colleges: 16 private institutions with admit rates above 25%
• More Selective Public Colleges: 10 public institutions with admit rates below 60%
• Selective Public Colleges: 9 public institutions with admit rates above 60%
ARC colleges in the More Selective Private and More Selective Public segments are quite
representative of all institutions in those same segments, while ARC colleges in the
Selective Private and Selective Public segments tend to be more selective than non-ARC
institutions in those same segments (see Appendix Table A1 for more detail).
ARC Data Definitions
This Brief employs the following terms and definitions:
Application cohorts, application cycles, and admission cycles are indexed according to
the fall entry term for which students applied (e.g., fall 2021). Because ARC studies the fall
2021 admissions process and its outcomes, all analyses in this report group students
according to the fall entry term for which they applied, even if students deferred first-year
enrollment to a later entry term. Thus, applicants for fall 2020 who deferred enrollment to fall
2021 are grouped as enrollees from the fall 2020 application cohort.
Applications refers to the aggregate number of applications ARC institutions collectively
received and evaluated, which is notably larger than the number of unique applicants who
applied to ARC institutions, since some students applied to several ARC institutions.
Admissions refers to the aggregate number of admissions offers ARC colleges collectively
extended, while admits and admitted students refer to unique students offered admission,
and admit rate refers to the fraction of applications offered admission.
Underrepresented Minority (URM) students are defined as students who are Black,
Hispanic/Latinx, Native American, Native Hawaiian/Pacific Islander, and Two or More Races.
Parental education and Income status data come from ARC colleges and are based on
the institutions’ own ways of collecting and designating parental education and which
students are low-income and not low-income.
Landscape neighborhood challenge is a measure of students’ neighborhood background,
used below as a proxy for students’ socioeconomic status. Landscape neighborhood
challenge is a neighborhood-level attribute constructed based on U.S. Census data and a
nationally representative sample of high school graduates, and it has been used by many
1
Admit rates are based on data from ARC colleges and are calculated as a weighted average using Fall 2018-2020 application
data.
s
11
participating colleges in recent application cycles. Neighborhood challenge levels are
expressed on a 1-100 percentile scale, where higher values indicate higher levels of
challenge related to educational opportunities and outcomes. For example, a neighborhood
with a Landscape challenge level of 64 has a higher level of educational challenge than 64%
of neighborhoods in the U.S. Likewise, 20% of U.S. neighborhoods fall into each quintile of
Landscape neighborhood challenge.
Thirty participating colleges provided data on applicants’ recalculated high school grade
point averages (HSGPAs). Because different colleges employ different grade scales when
recalculating applicant HSGPAs, we created five HSGPA quintiles using college-specific
HSGPA quintile cut-points based on the HSGPA distribution among each college’s fall 2018-
2020 applicants. At each college reporting recalculated HSGPAs, roughly 20% of fall 2018-
2020 applicants fall into each quintile of recalculated HSGPA.
When considering SAT/ACT scores, we distinguish between disclosed and recorded
SAT/ACT scores. Disclosed SAT/ACT scores are SAT/ACT scores that students submitted
to ARC colleges for consideration in the admissions process. Recorded SAT/ACT scores
include all SAT/ACT scores that students disclosed to ARC colleges as well as SAT scores
that applicants withheld but that are observable in College Board administrative data. For
students who did not disclose a score, the score utilized from College Board administrative
data is the highest combination of SAT section scores from across all of a student’s SAT
scores.
Feeder high schools are defined as high schools that sent more than 30 applications to an
institution over the fall 2018-2020 application cycles.
Throughout this Brief, data for groups of fewer than 10 observations are suppressed. In
tables, asterisks (*) denote data suppression for cells with fewer than 10 observations.
s
12
ARC Fall 2021 Admissions Cohort: Aggregate Changes in
College Applications, Admissions Offers, and Enrollment
Aggregate Enrollment Funnel Insights
Between the fall 2018 and fall 2021 admission cycles, ARC institutions received more than
5.1 million applications, extended nearly 2.3 million offers of admission, and enrolled nearly
650,000 first-year students. This sample comprises the dataset analyzed in this report.
2
Figure 1 displays counts of applications, admissions offers, and first-year enrollment at all
ARC institutions from fall 2018 through fall 2021. Applications sharply increased from fall
2020 to fall 2021, while admission offers steadily increased over the four years and
enrollment increased from fall 2020 to fall 2021 at higher rates than previous cycles.
Figure 1: ARC Colleges’ Enrollment Funnel, Fall 2018-2021
2
This analysis sample excludes about 1 million applications to ARC institutions that were incomplete or withdrawn before an
admission decision was reached.
s
13
Figure 2 compares funnel trends over time at all ARC institutions and the four segments of
ARC institutions. To better compare funnel trends across institution segments that differ in
size and selectivity, we express each data series as an index anchored to the fall 2018
academic year, which corresponds to the start of ARC historical data. Each index has a
value of 100 in the base year.
Key takeaways from Figure 2:
• Between the fall 2018 and fall 2021 application cycles, total applications to ARC
institutions increased 23%, but there is considerable variation across the four
segments, with More Selective Private ARC institutions experiencing the greatest
application growth (29%) over this three-year period.
• From fall 2018 to fall 2021, admission offers from ARC colleges grew by 20%.
Admissions growth occurred in all institutional segments except More Selective Private
colleges, where fall 2021 admissions offers declined by 4% compared to fall 2018.
• Across all ARC colleges, yield rates rebounded slightly in fall 2021 following a decrease
between fall 2019 and fall 2020. This rebound in yield rates occurred in all four
segments of ARC colleges and was strongest at More Selective Private ARC colleges.
• Between fall 2018 and fall 2021, enrollment at ARC colleges grew by 7%, with virtually
all of that growth occurring between fall 2020 and fall 2021.
Figure 2: Enrollment Funnel for ARC Colleges and Segments, Fall 2018-2021
s
14
Figure 3 shows aggregate changes in the funnel of all ARC colleges and each ARC
segment from the fall 2020 to fall 2021 cycle.
Key takeaways from Figure 3:
• Between fall 2020 and fall 2021, applications to ARC institutions grew by 17.8%. More
Selective Private ARC colleges experienced the largest application growth (29.4%).
• Between fall 2020 and fall 2021, admission offers from ARC institutions grew by 6.2%.
Offers grew by 10.4% at Selective Public colleges but declined by 12.5% at More
Selective Private colleges.
• In the aggregate, ARC institutions expanded first-year enrollment by 7.9% from fall
2020 to fall 2021. Enrollment grew substantially in all ARC segments except More
Selective Private ARC institutions, where enrollment increased 0.5% between fall 2020
and fall 2021.
Figure 3: Percentage Change in Applications, Admissions, and Enrollment Between Fall 2020 and Fall 2021,
ARC Colleges and Segments
s
15
Figure 4 shows how the 17.8% growth in application volume across all ARC institutions
breaks down across different student subgroups.
Key takeaways from Figure 4:
• Between fall 2020 and fall 2021, applications to ARC colleges grew among all student
subgroups.
• Application growth between fall 2020 and fall 2021 to ARC colleges was larger among
students with high HSGPAs than students with low HSGPAs.
• Application growth between fall 2020 and fall 2021 to ARC colleges was larger among
non-first generation students than first generation students.
• Application growth between fall 2020 and fall 2021 at ARC colleges was larger among
students from less challenging neighborhoods than more challenging neighborhoods.
Figure 4: Percentage Change in Applications to ARC Colleges Between Fall 2020 and Fall 2021, by Student
Characteristics
s
16
Table 1 provides data on percentage changes in applications between fall 2020 and fall
2021 for each of the four ARC segments alongside the aggregate ARC data presented in
Figure 4. Application growth between fall 2020 and fall 2021 was largest at More Selective
Privates, and substantial variations in application growth exist between segments by
race/ethnicity, parent education and low-income status.
Table 1: Percentage Changes in Applications Between Fall 2020 and Fall 2021, by ARC
Segment and Student Characteristics
Group
Percentage Change in Applications, Fall 2020-Fall 2021
ARC Colleges
Selective
Privates
All students
17.8%
16.9%
Gender
1
Female
20.0%
19.8%
Male
14.4%
13.0%
Race/Ethnicity
Asian
17.0%
17.1%
Black
17.2%
13.5%
Hispanic/Latinx
15.4%
17.7%
Native
13.3%
22.1%
Native Hawaiian/Pac. Islander
39.5%
36.5%
Two or More Races
21.6%
22.1%
White
16.9%
17.9%
Other
19.7%
13.7%
International
22.6%
14.6%
Parental education (44
colleges)
First-generation
15.9%
15.4%
Not first-generation
20.6%
17.5%
Parental education unknown
6.6%
8.7%
Income status (26 colleges)
Low-income
20.2%
-11.4%
Not low-income
19.4%
28.5%
Income unknown
15.7%
13.9%
Landscape NH challenge
Lowest NH challenge quintile
18.0%
18.6%
Second NH challenge quintile
19.1%
18.5%
Middle NH challenge quintile
16.6%
14.4%
Fourth NH challenge quintile
12.8%
12.2%
Highest NH challenge quintile
5.8%
4.8%
Missing NH challenge
23.9%
18.3%
Recalculated HSGPA (30
colleges)
HSGPA quintile 1 (lowest)
6.4%
9.0%
HSGPA quintile 2
4.9%
10.8%
HSGPA quintile 3
8.7%
14.6%
HSGPA quintile 4
11.3%
19.0%
HSGPA quintile 5 (highest)
25.2%
30.1%
HSGPA missing
24.9%
14.8%
1
Data for students of non-binary/other gender and unknown gender are omitted due to small sample sizes.
Note: Asterisks (*) denote data suppression for cells with <10 observations.
s
17
Figure 5 shows how the 6.2% growth in aggregate admission offers among ARC
institutions breaks down across different student subgroups.
Key takeaways from Figure 5:
• Between fall 2020 and fall 2021, aggregate admission offers to ARC colleges grew
among all demographic subgroups except international applicants and applicants with
lower high school grades.
• Admission offers growth at ARC colleges was larger among underrepresented minority
applicants than among Asian and White applicants.
• Admission offers growth was larger among students with high HSGPAs than among
students with low HSGPAs.
Figure 5: Percentage Change in Admission Offers at ARC Colleges Between Fall 2020 and Fall 2021, by Student
Characteristics
s
18
Table 2 provides data on growth in aggregate admission offers between fall 2020 and fall
2021 for each of the four ARC segments alongside the aggregate ARC college data
presented in Figure 5. Offers of admission grew among most student subgroups in all
segments except at More Selective Private ARC colleges, where offers of admission
primarily declined in fall 2021 relative to fall 2020 except among some minority student
subgroups, first generation, low-income and high challenge students.
Table 2: Percentage Changes in Admission Offers Between Fall 2020 and Fall 2021, by
ARC Segment and Student Characteristics
Group
Percentage Change in Admission Offers, Fall 2020-Fall 2021
ARC Colleges
Selective
Privates
All students
6.2%
5.7%
Gender
1
Female
9.2%
4.8%
Male
2.4%
6.9%
Race/Ethnicity
Asian
1.6%
3.7%
Black
17.1%
3.4%
Hispanic/Latinx
7.2%
2.1%
Native
16.5%
35.7%
Native Hawaiian/Pac. Islander
33.8%
68.4%
Two or More Races
11.4%
14.1%
White
6.6%
7.3%
Other
5.1%
-0.3%
International
-1.5%
3.2%
Parental education (44
colleges)
First-generation
4.2%
-4.7%
Not first-generation
5.5%
7.1%
Parental education unknown
10.2%
9.8%
Income status (26 colleges)
Low-income
0.4%
-21.5%
Not low-income
3.2%
14.9%
Income unknown
9.4%
3.3%
Landscape NH challenge
Lowest NH challenge quintile
5.7%
7.9%
Second NH challenge quintile
9.2%
5.9%
Middle NH challenge quintile
9.3%
-1.7%
Fourth NH challenge quintile
9.2%
-1.9%
Highest NH challenge quintile
6.5%
-9.3%
Missing NH challenge
-0.3%
8.0%
Recalculated HSGPA (30
colleges)
HSGPA quintile 1 (lowest)
-12.6%
-15.1%
HSGPA quintile 2
-6.9%
0.3%
HSGPA quintile 3
0.1%
9.7%
HSGPA quintile 4
3.0%
14.9%
HSGPA quintile 5 (highest)
15.7%
15.3%
HSGPA missing
10.6%
-1.8%
1
Data for students of non-binary/other gender and unknown gender are omitted due to small sample sizes.
Note: Asterisks (*) denote data suppression for cells with <10 observations.
s
19
Figure 6 shows how the 7.9% aggregate enrollment growth among ARC institutions breaks
down across different student subgroups.
Key takeaways from Figure 6:
• Between fall 2020 and fall 2021, enrollment at ARC colleges grew among all student
subgroups except students with lower high school grades.
• Enrollment growth was larger among students who were not first generation than
among first generation students, and larger among students who were not low-income
than among low-income students.
• Enrollment growth was slightly larger among students from moderately challenging
neighborhoods than among more-challenging and less-challenging neighborhoods.
Figure 6: Percentage Change in Enrollment at ARC Colleges Between Fall 2020 and Fall 2021, by Student
Characteristics
s
20
Table 3 provides data on aggregate enrollment growth between fall 2020 and fall 2021 for
each of the four ARC segments alongside the aggregate ARC college data presented in
Figure 6. Notably, enrollment growth at More Selective Private ARC colleges substantially
exceeds overall enrollment growth at ARC colleges among some minority student
subgroups, first generation, low-income, and high challenge students. Private college first-
year enrollments are roughly one-quarter the size of public colleges, on average, which is
consistent with larger percentage changes (see Appendix Table A2 for data on segment
sizes).
Table 3: Percentage Changes in Enrollment Between Fall 2020 and Fall 2021, by ARC
Segment and Student Characteristics
Group
Percentage Change in Enrollment, Fall 2020-Fall 2021
ARC Colleges
Selective
Privates
All students
7.9%
13.2%
Gender
1
Female
10.2%
13.8%
Male
5.2%
12.4%
Race/Ethnicity
Asian
7.7%
15.5%
Black
9.6%
16.0%
Hispanic/Latinx
9.8%
10.4%
Native
26.3%
72.7%
Native Hawaiian/Pac. Islander
20.0%
64.7%
Two or More Races
12.7%
10.5%
White
6.7%
13.8%
Other
9.1%
15.2%
International
8.6%
8.3%
Parental education (44
colleges)
First-generation
7.1%
10.8%
Not first-generation
9.4%
14.6%
Parental education unknown
3.1%
-4.1%
Income status (26 colleges)
Low-income
7.9%
2.0%
Not low-income
9.5%
29.2%
Income unknown
6.4%
6.5%
Landscape NH challenge
Lowest NH challenge quintile
6.9%
14.1%
Second NH challenge quintile
10.3%
15.5%
Middle NH challenge quintile
9.3%
10.3%
Fourth NH challenge quintile
8.7%
9.3%
Highest NH challenge quintile
6.6%
6.0%
Missing NH challenge
8.9%
11.1%
Recalculated HSGPA (30
colleges)
HSGPA quintile 1 (lowest)
-8.6%
-9.7%
HSGPA quintile 2
-4.0%
-1.1%
HSGPA quintile 3
4.1%
17.4%
HSGPA quintile 4
9.2%
20.3%
HSGPA quintile 5 (highest)
22.7%
19.4%
HSGPA missing
8.4%
15.2%
1
Data for students of non-binary/other gender and unknown gender are omitted due to small sample sizes.
Note: Asterisks (*) denote data suppression for cells with <10 observations.
s
21
Aggregate Insights into Racial and Economic Diversity
Figure 7 shows how the racial/ethnic composition of ARC colleges’ first-year enrollees
changed between fall 2018 and fall 2021.
3
• Across all ARC colleges, the share of underrepresented minority (URM) enrollees grew
steadily from 23.9% in fall 2018 to 25.4% in fall 2021, increasing by 0.5 percentage
points with each admission cycle.
Figure 7: Racial/Ethnic Composition of Enrolled Students from Fall 2018 to Fall 2021, ARC Colleges
3
See Appendix Figures 28-31 for changes in the racial/ethnic composition of first-year enrollees in each ARC segment
between Fall 2018 and Fall 2021.
23.9%
24.4%
24.9%
25.4%
s
22
Focusing on the past two admission cycles, Figure 8 shows how the racial/ethnic
composition of enrolled students changed between fall 2020 and fall 2021 at ARC colleges
and within each college segment. Despite growth in URM student enrollment, the
racial/ethnic composition of first-year enrollees changed minimally between fall 2020 and fall
2021 at ARC colleges because enrollment among non-URM students grew at a similar rate.
• Across all ARC colleges, the URM share of first-year enrollees increased from 24.9% in
fall 2020 to 25.4% in fall 2021, consistent with 0.5 percentage point growth in the URM
share of first-year enrollees over each of the last four years.
• More Selective Private ARC colleges’ first-year enrollees were 26.0% URM students in
fall 2020 and grew to 29.4% URM students in fall 2021, but the URM share of enrollees
remained roughly flat in the other segments.
• Considerable variation exists across ARC institutions, with roughly half of institutions
experiencing increases and the other half experiencing decreases in URM
representation between fall 2020 and fall 2021.
• The patterns observed below – nearly flat URM student representation in ARC
segments except for More Selective Private ARC institutions – is consistent with
analyses based on a near-universal set of U.S. four-year institutions using College
Board and National Student Clearinghouse data (Howell et al., 2021, 2022).
Figure 8: Racial/Ethnic Composition of Enrolled Students in Fall 2020 and Fall 2021, ARC Colleges and ARC
Segments
s
23
Figure 9 uses Landscape neighborhood challenge data to show how the socioeconomic
composition of ARC colleges’ first-year enrollees changed between fall 2018 and fall 2021.
4
• Across all ARC colleges, the share of domestic first-year enrollees from neighborhoods
in the three highest challenge quintiles remained quite stable, declining modestly from
23.1% in fall 2018 to 22.9% in fall 2021.
Figure 9: Socioeconomic Composition of Enrolled Students from Fall 2018 to Fall 2021, ARC Colleges
4
Figure 9 excludes students missing Landscape neighborhood challenge information. Since Landscape challenge data are not
available for international students, Figure 9 represents the socioeconomic composition of domestic enrollees by
neighborhood challenge quintile. See https://secure-media.collegeboard.org/landscape/comprehensive-data-methodology-
overview.pdf for more information on neighborhood challenge data.
s
24
Focusing on the past two admission cycles, Figure 10 shows how the socioeconomic
composition of enrolled students changed between fall 2020 and fall 2021 at ARC Colleges
and within each college segment.
5
Despite growth in the number of first generation, low
income, and higher-challenge students, the socioeconomic composition of enrollees in most
ARC segments changed minimally between fall 2020 and fall 2021.
• Across all ARC colleges, the share of students from the fourth and fifth neighborhood
challenge quintiles was 12.5% in both fall 2020 and fall 2021.
• Only More Selective Private ARC colleges experienced an increase in the share of
students from more challenging neighborhoods between fall 2020 and fall 2021.
• Between fall 2020 and fall 2021, roughly half of ARC colleges experienced increases in
the mean neighborhood challenge level of their first-year enrollees, while the other half
experienced decreases in first-year enrollees’ mean neighborhood challenge level.
• The patterns below in ARC segments also hold in the larger population of colleges and
universities based on analyses of National Student Clearinghouse data.
Figure 10: Socioeconomic Composition of Enrolled Students in Fall 2020 and Fall 2021, ARC Colleges and ARC
Segments
5
Figure 10 excludes students missing Landscape neighborhood challenge information. Since Landscape challenge data are
not available for international students, Figure 10 represents the socioeconomic composition of domestic enrollees by
neighborhood challenge quintile. See https://secure-media.collegeboard.org/landscape/comprehensive-data-methodology-
overview.pdf for more information on neighborhood challenge data.
s
25
Aggregate Insights into Test Score Disclosure and Withholding
Linking ARC institutions’ applicant records to College Board data enables us to distinguish
fall 2021 applicants with no recorded SAT score, applicants who disclosed scores, and
applicants who withheld SAT scores. Specifically, we categorize fall 2021 applicants into
three groups:
1. No-SAT applicants had no SAT score and either withheld an ACT score or had no ACT
score.
6
2. Score disclosers are applicants who provided an SAT/ACT score for consideration in
admissions.
3. Score withholders are applicants who had an SAT score and withheld it.
No-SAT score applicants and score withholders are indistinguishable to institutions in the
sense that neither group presents a test score in their application, but these two groups are
distinguishable using College Board data.
Those applicants to ARC colleges who had no test score likely had fewer opportunities to
test because of closed schools during the pandemic. At most, 23% of applications to ARC
colleges had no college entrance exam. This is an upper bound because these students had
no SAT score but may have had an ACT score that they withheld from their college
application, but we cannot observe withheld ACT scores in this data.
No-SAT applicants were much more likely to attend high schools where there were
substantial declines in SAT-taking relative to the pre-pandemic year, likely the result of their
high school being closed. While students may have been able to take an SAT at a high
school other than their own, we use this as a proxy for reduced opportunity to test. We also
know that no-SAT applicants to ARC colleges are more likely to live in certain regions of the
U.S. – West Coast, Mid-Atlantic, and New England – a pattern that is also associated with
known geographic differences in where schools were closed and closed for longer. Finally,
no-SAT applicants to ARC colleges are more likely to be international students where there
were pandemic-related disruptions to testing opportunities as well. Covid-related access to
testing appears to have been a transitory disruption primarily felt by the class of 2021.
6
We cannot distinguish ACT withholders from applicants who had no ACT score. As a result, our count of score withholders is
a lower bound on the true number of score withholders; conversely, our count of no-score applicants is an upper bound.
s
26
Figure 11 shows the distribution of fall 2021 applicants by score disclosure status in each
college segment and at all ARC institutions.
• Across all applications to ARC institutions, at most 23.4% of applications came from
applicants with no test score, 48.2% of applications disclosed an SAT/ACT score, and
at least 28.4% came from applicants who had a test score but withheld it.
• Test score disclosure rates are substantially higher among applications to More
Selective Public ARC institutions.
Figure 11: Score Disclosure, Withholding, and Absence Among Fall 2021 Applicants, ARC Colleges and
Segments
s
27
Ordinary Least Squares (OLS) regression is the most common statistical method used to
understand what factors influence an outcome. The outcome examined in this case is test
score disclosure. We use OLS to estimate the variation in test score disclosure decisions
that is explained by data represented in Figure 12 and find that SAT score is the strongest
predictor of a student’s decision to disclose a test score. Figure 12 shows the strong
correlation between a student’s SAT score (relative to the college they are applying to) and
the probability of disclosing a score on their application for fall 2021.
Examining combinations of predictors, like SAT score and HSGPA, further reveals that
HSGPA adds minimal incremental predictive strength in determining who will disclose a test
score. Continuing to add predictors that are student demographic characteristics, like
race/ethnicity and parental education, adds minimal to no incremental predictive strength.
7
Figure 12: Predictors of Test Score Disclosure for Fall 2021
7
The simple (single-variable) correlations with test score disclosure are SAT (0.519), Race/Ethnicity (0.204), HSGPA (0.151),
and Parental Education (0.140). SAT enters the model first because it has the strongest simple correlation. Although
variables could be added in order of their simple correlative strength, it is common to include all academic achievement
variables (e.g., SAT and HSGPA) before adding demographic variables (e.g., Race/Ethnicity and Parental Education) to
better understand the predictive strength of including academics and demographics. Results are not sensitive to the order
in which variables are added to the model. Additional predictors in the final model include feeder high school status, in-
state status, neighborhood and high school challenge, and college-specific indicators.
s
28
Results from the fully specified model with all predictors and the most predictive strength
(see Appendix for regression details) can be displayed in an alternate way. Figure 13 shows
predicted probabilities of test score disclosure for applicants with different test scores
relative to the college to which they applied.
8
For example, a student with a 30th percentile
SAT score relative to the college average has a score that is in the 30th percentile
compared to the distribution of applicants’ scores at the ARC college to which they applied.
9
The steepness of the line in Figure 13 demonstrates the strong relationship between relative
test score and the disclosure decision identified in Figure 12. Applicants with relatively
low test scores have a low predicted probability of disclosing those scores;
applicants with relatively high test scores have a high predicted probability of
disclosing those scores.
Figure 13: Probability of Test Score Disclosure Among ARC College Applications for Fall 2021
8
Probabilities in Figures 13 – 16 are based on a logistic regression model that also controls for race, in-state status, first-gen
status, feeder school status, HSGPA, neighborhood and high school challenge, and indicators for the sampled colleges.
Very similar patterns emerge in un-modeled data on score disclosure at every ARC institution.
9
The college average SAT score is calculated using data on all SAT and concorded ACT scores from 2018 to 2021 cohorts.
s
29
Plotting score disclosure probability curves for applicants with different HSGPAs in Figure 14
reveals small differences in disclosure behavior among applicants with different high school
grades. For example, ARC college applications with high school grades of B+ are always
somewhat more likely to disclose a test score compared to applications with grades of A+
and A.
10
Figure 14: Probability of Test Score Disclosure Among ARC College Applications for Fall 2021, by HSGPA
Figure 15 shows score disclosure probability curves by parental education. Figure 15
demonstrates that first generation and non-first generation applicants with the same relative
test score have virtually identical probabilities of disclosing that score to ARC colleges.
Figure 16 shows score disclosure probability curves by race/ethnicity. Figure 16
demonstrates small differences in disclosure behavior among applicants from different
racial/ethnic subgroups. For example, among applications with low relative scores (e.g.,
20th percentile), all racial/ethnic subgroups share a low probability of disclosing their scores.
10
High school grades below B+ are not displayed because 90% of applications to ARC colleges have HSGPAs of B+ or higher.
s
30
Figure 15: Probability of Test Score Disclosure Among ARC College Applications for Fall 2021, by Parental
Education
Figure 16: Probability of Test Score Disclosure Among ARC College Applications for Fall 2021, by Race/Ethnicity
s
31
Practically, the regression results presented graphically in Figures 13-16 demonstrate that
students with the same test score and high school grades are predicted to make very similar
score disclosure decisions when applying to ARC colleges regardless of whether they are
URM or non-URM, first generation or non-first generation, or low-income or not low-income.
Regression analyses reveal that test score, relative to test scores at the college to which
students apply, is the strongest determinant of a student’s decision to disclose a score,
where higher scoring students disclose scores at higher rates than lower scoring students,
on average. High school grades and student demographics add little additional power to
predict score disclosure decisions. Previously published differences in score disclosure
patterns by race, parental education, and income are attributable to differences in academic
achievement among score disclosers and withholders, for which prior analyses were unable
to control (Freeman et al., 2021).
s
32
Conclusion
The ARC initiative enabled a unique set of data and analyses to better understand changes
in college and student behavior influenced by the covid-19 pandemic. This Research Brief
provides initial evidence based on more than 50 ARC colleges on changes in application,
admission, and enrollment trends in the fall 2021 college application cycle compared to prior
years. The analyses also begin to explain students’ decisions to disclose or withhold their
standardized test score in the college application process.
Three key themes emerged from the initial analyses of the fall 2021 college application
cycle:
1. Between fall 2020 and fall 2021, ARC colleges experienced increases in the number of
applications (17.8%), offers of admission (6.2%), and students enrolled (7.9%) at rates
above and beyond prior years. At ARC colleges, the increases in applications, offers of
admission, and enrollment between fall 2020 and fall 2021 were experienced by nearly
every student subgroup.
2. Because enrollment of URM students and non-URM students increased at ARC
colleges, the representation of URM students changed very little. The exception is
More Selective Private ARC colleges, where the share of Black, Hispanic, Native, and
Two or More Races students increased between fall 2020 and fall 2021. The
proportional representation of lower-socioeconomic status students was either
unchanged or decreased at ARC colleges.
3. At most, about 20% of ARC applications in fall 2021 had no recorded SAT/ACT score.
Among the other 80% of ARC applications, approximately 50% disclosed an SAT/ACT
score and 30% withheld a score. Students’ decisions to withhold or disclose their test
score is nearly entirely explained by the test score relative to the college to which they
apply – students with high relative scores tend to disclose them, while students with low
relative scores tend to withhold them. Students with the same test scores and high
school grades applying to the same college are predicted to make very similar score
disclosure decisions regardless of demographics like race/ethnicity and socioeconomic
status.
ARC colleges will continue to share data on student performance to enable continued
research on how pandemic-related disruptions affect student outcomes like college grades,
credit accumulation, and retention. Future Research Briefs will share those analyses as well
as replicate the analyses in this Brief for the next cohort of applicants to ARC colleges.
s
33
Appendix
Table A1: Attributes of Colleges in ARC and Each Institutional Segment
More Selective
Private Colleges
Selective Private
Colleges
More Selective
Public Colleges
Selective Public
Colleges
ARC
Non-
ARC
ARC
Non-
ARC
ARC
Non-
ARC
ARC
Non-
ARC
Number of Institutions
16
37
16
766
10
79
9
399
First-Year Enrollment
1,682
969
1,197
471
5,928
2,220
5,466
1,778
Admit Rate
14%
14%
44%
72%
48%
45%
71%
82%
Yield Rate
43%
45%
20%
22%
30%
31%
23%
27%
First-Year Pell Share
17%
18%
15%
39%
26%
43%
21%
42%
Percent In-State
16%
23%
33%
60%
77%
76%
57%
82%
Tuition and Fees
$58,114
$54,557
$51,538
$34,668
$12,486
$9,657
$13,646
$9,782
Graduation Rate
93%
90%
84%
59%
80%
63%
77%
53%
First-Year
Racial/Ethnic
Composition
Native American
0%
0%
0%
0%
0%
0%
0%
1%
Asian American
16%
16%
9%
3%
11%
10%
10%
4%
Black/African
American
7%
6%
5%
13%
12%
19%
5%
14%
Hispanic
12%
12%
11%
12%
12%
19%
11%
14%
Native Hawaiian
0%
0%
0%
0%
0%
0%
0%
0%
White
45%
47%
60%
1%
54%
42%
61%
57%
Two Or More Races
6%
6%
5%
4%
4%
4%
4%
4%
Unknown
2%
3%
2%
6%
2%
2%
2%
3%
International
11%
10%
9%
4%
4%
4%
7%
2%
Sources: IPEDS 2020 Survey, ARC data
Notes: More Selective Private Colleges have admit rates less than or equal to 25%. More Selective Public
Colleges have admit rates less than or equal to 60%.
s
34
Table A2: Number of Applications, Admissions Offers, and Enrollments, by Fall Application
Cohort and ARC Institutional Segment
Fall Application Cohort
2018
2019
2020
2021
All ARC Colleges
Number of Applications
1,192,635
1,251,164
1,245,915
1,467,999
Number of Admissions Offers
528,618
534,978
598,573
635,808
Number of Enrollments
158,430
156,949
156,787
169,239
More Selective Private ARC Colleges
Number of Applications
380,250
400,638
380,372
492,288
Number of Admissions Offers
59,716
55,547
65,766
57,565
Number of Enrollments
26,665
25,794
26,927
27,069
Selective Private ARC Colleges
Number of Applications
210,910
213,925
208,896
244,262
Number of Admissions Offers
97,684
93,496
103,905
109,786
Number of Enrollments
20,656
19,937
19,734
22,334
More Selective Public ARC Colleges
Number of Applications
341,694
356,688
361,952
407,653
Number of Admissions Offers
173,918
179,582
203,130
219,271
Number of Enrollments
58,732
58,045
59,309
63,273
Selective Public ARC Colleges
Number of Applications
259,781
279,913
294,695
323,796
Number of Admissions Offers
197,300
206,353
225,772
249,186
Number of Enrollments
52,377
53,173
50,817
56,563
Source: ARC data.
Note: There are 51 ARC colleges in the full sample; 16 More Selective Private ARC colleges (average first-year
enrollment of 1,692 students), 16 Selective Private ARC colleges (average first-year enrollment of 1,396
students), 10 More Selective Public ARC colleges (average first-year enrollment of 6,327 students), and 9
Selective Public ARC colleges (average first-year enrollment of 6,285 students).
s
35
Test Score Disclosure Regression Details
In order to estimate the how the probability of SAT score disclosure (

) varies with a
student’s SAT score, we fit the ARC data with the logistic regression model expressed
through Equation (1). In this model, the following variables are defined as:
• SATScore is a student’s actual SAT score inclusive of disclosed and withheld scores.
Disclosed scores are sourced from ARC colleges, while withheld scores are the highest
combination of SAT section scores from across all of a student’s SAT scores in College
Board administrative data.
• Race is a vector of indicator variables expressing the student race/ethnicity provided by
the college.
• ParentalEd is a vector of indicator variables expressing the student’s parental
education from the College Board’s Student Data Questionnaire (SDQ).
• InState is an indicator for whether the student resides in the same state as the college
to which they applied.
• HSGPA is the student’s self-reported high school GPA from the College Board’s
Student Data Questionnaire (SDQ) on a 0-4.33 scale.
• HSChallenge and NHChallenge express the student’s high school and neighborhood
challenge on a 1-100 scale, where higher challenge levels indicate more
disadvantaged neighborhoods and high schools.
• Feeder is an indicator variable identifying students that sent 30 or more applications to
the college between 2018 and 2020.
• College is a vector of college fixed effects to capture differences across ARC
institutions that are constant for all students.
In equation (1), parameter identifies the interaction of the Race indicator variables and the
student’s SAT score. This interaction allows for different relationships between the student’s
SAT score and disclosure probability for different subgroups of students, defined by race. In
models where we identify the relationships between HSGPA or Parental Education and SAT
score disclosure, parameter expresses the interaction of these variables and SAT score.
To construct the fitted score disclosure curves depicted in Figures 13-16, we hold all
variables constant at their sample means and use parameters , , and to demonstrate
how SAT disclosure probabilities change with SAT scores for different subgroups of
students.
Equation (1):














s
36
References
Freeman, M., P. Magouirk, and T. Kajikawa. 2021. Applying to college in a test-optional
admissions landscape: trends from Common App data. Common Application: Arlington, VA.
September 8, 2021. Retrieved from https://s3.us-west-
2.amazonaws.com/ca.research.publish/Research+briefs+2020/20210908_Paper4_TestOpti
onal.pdf.
Howell, J., M. Hurwitz, J. Ma, G. Perfetto, J. Wyatt, and L. Young. 2021. College enrollment
and retention in the era of covid: Fall 2020. College Board: New York, NY. June 2021.
Retrieved from https://research.collegeboard.org/media/pdf/enrollment-retention-
covid2020.pdf.
Howell, J., M. Hurwitz, J. Ma, G. Perfetto, J. Wyatt, and L. Young. 2022. College enrollment
and retention in the era of covid: Fall 2021. College Board: New York, NY. Forthcoming
2022. Available at https://research.collegeboard.org/.
About the College Board
The College Board is a mission-driven not-for-profit organization that connects students to
college success and opportunity. Founded in 1900, the College Board was created to
expand access to higher education. Today, the membership association is made up of over
6,000 of the world’s leading educational institutions and is dedicated to promoting
excellence and equity in education. Each year, the College Board helps more than seven
million students prepare for a successful transition to college through programs and services
in college readiness and college success — including the SAT
®
and the Advanced
Placement Program
®
. The organization also serves the education community through
research and advocacy on behalf of students, educators and schools. For further
information, visit www.collegeboard.org.