Assessing a New Approach to Class-Based Affirmative Action

183 Pages·2011·1.2 MB·English

i

ASSESSING A NEW APPROACH TO CLASS-BASED AFFIRMATIVE ACTION
by
MATTHEW NEWMAN GAERTNER
B.A., Georgetown University, 2000

A dissertation submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
School of Education
Research and Evaluation Methodology
2011

This material is based upon work supported by the Association for Institutional Research, the
National Center for Education Statistics, the National Science Foundation, and the National
Postsecondary Education Cooperative, under Association for Institutional Research Grant
Number DG10-206. Any opinions, findings, and conclusions or recommendations expressed in
this material are those of the author and do not necessarily reflect the views of the Association
for Institutional Research, the National Center for Education Statistics, the National Science
Foundation, or the National Postsecondary Education Cooperative.
ii

This dissertation entitled:
Assessing a New Approach to Class-Based Affirmative Action
written by Matthew Newman Gaertner
has been approved for the School of Education, Department of Research and Evaluation
Methodology

__________________________________________
Derek Briggs, Ph.D. (co-chair)

__________________________________________
Edward Wiley, Ph.D. (co-chair)

__________________________________________
Michele Moses, Ph.D.

__________________________________________
Lorrie Shepard, Ph.D.

__________________________________________
Melissa Hart, J.D.

Date ______________

The final copy of this thesis has been examined by the signatories, and we find that both the
content and the form meet acceptable presentation standards of scholarly work in the above
mentioned discipline.
IRB Protocol # 1209.5
iii

Gaertner, Matthew Newman (Ph.D., Education, Research and Evaluation Methodolgy)
Assessing a New Approach to Class-Based Affirmative Action
Dissertation directed by Derek Briggs, Ph.D., and Edward Wiley, Ph.D.

ABSTRACT
In November, 2008, Colorado and Nebraska voted on amendments that sought to end
race-based affirmative action at public universities in those states. In anticipation of the vote, the
University of Colorado at Boulder (CU) explored statistical approaches to support class-based
(i.e., socioeconomic) affirmative action. This dissertation introduces CU‟s method of identifying
socioeconomically disadvantaged and overachieving applicants in undergraduate admissions. In
addition, sensitivity analyses were conducted to gauge the impact of technical decisions that
were made when these measures were devised. Two experiments were carried out to determine
whether or not implementing this approach would change the racial and socioeconomic diversity
of accepted classes. Finally, historical student records were examined to explore the likelihood of
college success for the beneficiaries of CU‟s class-based approach.
The sensitivity analyses identify particularly consequential issues that architects of class-
based systems may face, including modeling application to college, defining target populations,
and addressing missing data. The experiments suggest class-based affirmative action can
potentially increase acceptance rates for low-SES and minority applicants, particularly if it is
used alongside race-conscious admissions. Analyses of historical data do not rule out the
possibility of college success for the beneficiaries of class-conscious admissions, but they do
argue for the provision of robust academic support to marginally qualified, low-SES students
when they matriculate. This dissertation is intended to serve as a resource for postsecondary
iv

institutions considering class-based admissions policies. If race-based approaches are overturned,
universities like CU could struggle to develop race-blind metrics to identify applicants who have
faced adversity. This research examines one method of quantifying the barriers these students
encounter.
v

DEDICATION
For Freya and Corinne
vi

ACKNOWLEDGEMENTS
Over the past five years I have had the privilege of learning from and collaborating with
thoughtful and dedicated scholars at the University of Colorado at Boulder. First, to my co-
chairs, mentors, and friends, Derek Briggs and Ed Wiley. Their tireless commitment to my
development as a researcher was instrumental at every stage of this dissertation. I would like to
thank them both for their generosity, support, and good humor. Derek and Ed have been an
inspiration; without their efforts, I surely could not have come this far.
I am also indebted to my energetic and enthusiastic dissertation committee – Michele
Moses, Lorrie Shepard, and Melissa Hart. Each provided me with helpful and carefully
considered feedback throughout my doctoral career. Thank you all.
Special thanks and recognition is owed to my colleagues at CU‟s Office of Admissions.
In particular, I would like to recognize Alex Subert, Kevin MacLennan, and Matt Lopez. This
dissertation would not have been possible without their active participation. From conception
through completion, Alex, Kevin, and Matt have championed this work and provided every
resource necessary for its success.
I would also like to thank my colleagues in the School of Education for the years of
friendship, laughs, and encouragement: Magda Chia, Nathan Dadey, Ben Domingue, Amy
Farley, Ellie Fulbeck, Adam Van Iwaarden, and Jon Weeks.
Most importantly, I am grateful for my loving family – Karen, Greg, David, Linda, Paul,
and Nikki. Their support throughout graduate school deserves my deepest thanks. To my wife
Freya and daughter Corinne: Your love, humor, patience, feedback (Freya), technical skills
(Corinne), and dancing prowess (Freya & Corinne) made this journey possible. Thank you.
vii

CONTENTS
Page
ABSTRACT iii
DEDICATION v
ACKNOWLEDGEMENTS vi
LIST OF TABLES x
LIST OF FIGURES xii
CHAPTER
1. INTRODUCTION 1
Research Problem 2
Expected Contribution 7
Chapter Overview and Research Questions 8
2. BACKGROUND AND LITERATURE REVIEW 11
A Brief History of Affirmative Action 12
The Rise of Class-Based Affirmative Action 15
3. UNDERGRADUATE ADMISSIONS AT THE UNIVERSITY OF COLORADO AT
BOULDER 25
Application, Acceptance, and Enrollment 26
Minimum Standards for Admissibility 30
Primary and Secondary Factors in the Admissions Process 32
Class-Based Considerations in the Admissions Process 38
Class-Based Measures as Primary and Secondary Factors 38
Relevant Qualifications 39
viii

4. DEVELOPING MEASURES TO SUPPORT CLASS-BASED AFFIRMATIVE
ACTION 43
The Disadvantage and Overachievement Indices 44
The Disadvantage Index 45
The Overachievement Index 51
Implementing the Indices at CU 57
Data Sources 57
ELS Variables 58
Establishing Thresholds along the Indices‟ Scales 60
Considering the Indices in Admissions Decisions 64
5. SENSITIVITY OF THE INDICES TO THE ELS SAMPLE AND REGRESSION
MODELS 67
Methods 70
Findings 74
Relaxing Restrictions on the ELS Dataset 75
Modeling Application to College 83
Imputing School-Wide Poverty 90
Adding a School-Level College-Going Measure 95
Comprehensively Revised ELS Samples and Indices Models 97
6. IMPACT OF CLASS-BASED APPROACH ON ACCEPTANCE RATES 99
Comparing Race-Based and Class-Based Affirmative Action 100
Methods 100
Findings 102
Comparing Race-Based and Class-Plus-Race Affirmative Action 107
Methods 107
ix

Findings 111
Examining Sample Attrition 117
7. COLLEGE OUTCOMES FOR CLASS-BASED ADMITS 120
Methods 121
Coarsened Exact Matching 123
Findings 127
Choosing and Coarsening Covariates 127
Developing Weights 128
Evaluating Balance 129
Outcomes for Class-Based Admits 131
8. DISCUSSION 137
Implications for Class-Based Affirmative Action 138
Limitations 143
Future Research 146
A Final Note on Class-Conscious Admissions 149
REFERENCES 151
APPENDICES 161
Chapter 3 Appendices 162
Chapter 5 Appendices 163
x

TABLES
Table
4.1 Socioeconomic and Academic Predictors for the Disadvantage Index 46
4.2 Socioeconomic Characteristics of the Typical CU Applicant 47
4.3 Parameter Estimates for CU‟s Model of College Enrollment 49
4.4 Socioeconomic Predictors for the Overachievement Indices 52
4.5 Parameter Estimates for CU‟s Models of HSGPA, ACT scores, and SAT scores 55
4.6 Disadvantage and Overachievement Thresholds for the Indices 63
4.7 Additional Consideration Granted to Disadvantaged and Overachieving Applicants 64
5.1 Analytic Focus, by Sensitivity Analysis 73
5.2 Changes in Classifications for Additional Consideration, by Sensitivity Analysis 74
5.3 Changes in Classifications for the Overachievement Index (HSGPA), Following the
Removal of Restrictions to ELS Sample 80
5.4 Changes in Classifications for Additional Consideration, Following the Removal of
ELS Sample Restrictions 81
5.5 Changes in Classifications for the Disadvantage Index, Following the Removal of
Restrictions to ELS Sample 81
5.6 Changes in Variance Explained Following Imputation of %FRL 94
6.1 Acceptance Rates by Admissions Condition and Subgroup 102
6.2 High School Academic Credentials of Applicants, by Experimental Condition and
Admissions Decision, 2009 Experiment 104
6.3 Comparison of Academic Credentials by Category of Admissions Decision, 2009
Experiment 106
6.4 Acceptance Rates by Admissions Condition for Low-SES Applicants, 2010
Experiment 112
6.5 Acceptance Rates by Admissions Condition for Minority Applicants, 2010
Experiment 113