Selection Bias in Educational Debt Decisions: Analyzing the Impacts

Selection Bias in Educational Debt Decisions:
Analyzing the Impacts on Enrollment in Master’s
Degree Programs
Alee Lynch-Gunderson, PhD Student
Dr. Pete Villarreal III, Faculty
University of Florida
School of Human Development and Organizational Studies
Higher Education Administration
Selected Citations
 Millett (2003) “How Undergraduate Loan Debt Affects
Application and Enrollment in Graduate or First
Professional School”
 Purpose: Effects of debts on students who are most
likely prospects for entering graduate and professional
school
 Methodology: Logistic regressions
 Sample: Recent bachelor’s degree recipients who expect
to earn a doctoral degree (1,982 cases)
 Dataset: Baccalaureate and Beyond: 93/97
 Key Limitation: Dataset did not distinguish types of
financial support
Selected Citations
 Malcolm & Dowd (forthcoming) “College Student Debt as
Opportunity or Disadvantage? A Reconceptualization and
Application to STEM Graduate Enrollment”
 Purpose: Effect of debt on graduate school attendance of




STEM majors
Methodology: Propensity Score Matching
Sample: STEM Bachelor’s recipients from 2000-01 & 2001-02
(7,700 cases)
Dataset: 2003 National Survey of Recent College Graduates,
2002-2003 College Board Annual Survey of Colleges and
Universities, Institute for College Access and Success,
Integrated Postsecondary Education Data System, & Barron’s
Profiles of American Colleges
Key Limitation: Exclusion of other master’s degree programs
Selected Citations
 Perna (2004) “Understanding the Decision to Enroll in
Graduate School: Sex and Racial/Ethnic Group Differences”
 Purposes:
 Contribute to understanding of underrepresentation of women,
African Americans, and Hispanics among doctoral and professional
degree enrollees
 Test a conceptual model for graduate school enrollment
 Method: Multinomial logit models
 Sample: Bachelor’s degree recipients in 1992-93 (9,241 cases)
 Dataset: Baccalaureate and Beyond: 93/97
 Key Limitation: Did not control for self-selection bias
Sex
• Male
• Female
Cultural & Social Capital
• Parental educational attainment
• Primary language at home is English
• Values additional education (B&B 11item)
• Parental monetary contribution
• Carnegie classification
• Tuition
• Location
• Attended two-year college
Financial & Academic
Resources
• Undergraduate Educational Debt
• Dependency status
• Time to Bachelor’s Degree
• Cumulative Undergraduate GPA
• SAT/ACT quartile
Race/Ethnicity
Perna’s
Enrollment
Decision
Conceptual
Model
• Asian
• Black
• Hispanic
• White
• other
Expected Costs & Benefits
• Net price
• Foregone earnings by undergraduate
major
• Time horizon (delayed college)
• Marital Status
• Parental Status
Cultural & Social Capital
•Parental educational attainment
•Primary language at home is English
•Values additional education (B&B
11-item)
•Parental monetary contribution
•Carnegie classification
•Tuition
•Location
•Attended two-year college
Sex
• Male
• Female
Cultural & Social Capital
• Parental educational attainment
• Primary language at home is English
• Values additional education (B&B 11item)
• Parental monetary contribution
• Carnegie classification
• Tuition
• Location
• Attended two-year college
Financial & Academic
Resources
• Undergraduate Educational Debt
• Dependency status
• Wen bachelor’s received
• Cumulative Undergraduate GPA
• SAT/ACT quartile
Race/Ethnicity
Perna’s
Enrollment
Decision
Conceptual
Model
• Asian
• Black
• Hispanic
• White
• other
Expected Costs & Benefits
• Net price
• Foregone earnings by undergraduate
major
• Time horizon (delayed college)
• Marital Status
• Parental Status
Expected Costs & Benefits
•Net price
•Foregone earnings by undergraduate
major
•Time horizon (delayed college)
•Marital Status
•Parental Status
Sex
• Male
• Female
Cultural & Social Capital
• Parental educational attainment
• Primary language at home is English
• Values additional education (B&B 11item)
• Parental monetary contribution
• Carnegie classification
• Tuition
• Location
• Attended two-year college
Financial & Academic
Resources
• Undergraduate Educational Debt
• Dependency status
• Wen bachelor’s received
• Cumulative Undergraduate GPA
• SAT/ACT quartile
Race/Ethnicity
Perna’s
Enrollment
Decision
Conceptual
Model
• Asian
• Black
• Hispanic
• White
• other
Expected Costs & Benefits
• Net price
• Foregone earnings by undergraduate
major
• Time horizon (delayed college)
• Marital Status
• Parental Status
Financial & Academic Resources
•Undergraduate Educational Debt
•Dependency status
•Time to Bachelor’s Degree
•Cumulative Undergraduate GPA
•SAT/ACT quartile
Student Characteristics
• Sex
• Race/Ethnicity
Cultural & Social Capital
• Parental educational attainment
• Primary language at home is English
• Values additional education (B&B 11item)
• Parental monetary contribution
• Carnegie classification
• Tuition
• Location
• Attended two-year college
• Parent’s have a mortgage
• Location of employment
Expected Costs & Benefits
• Net price
• Foregone earnings by undergraduate
major
• Time horizon (delayed college)
• Marital Status
• Parental Status
Modified
Enrollment
Decision
Conceptual
Model
Academic Resources
Financial Resources
• Time to Bachelor’s Degree
• Cumulative Undergraduate GPA
• SAT/ACT quartile
• Undergraduate Educational Debt
• Dependency status
• Type of Assistantship
Financial Resources
•Undergraduate Educational Debt
•Dependency status
•Type of Assistantship
Academic Resources
•Time to Bachelor’s Degree
•Cumulative Undergraduate GPA
•SAT/ACT quartile
Cultural & Social Capital
•Parental educational attainment
•Primary language at home is English
•Values additional education (B&B 11-item)
•Parental monetary contribution
•Carnegie classification
•Tuition
•Location
•Attended two-year college
•Parent’s have a mortgage
•Location of employment
Research Study
 How does the likelihood of master’s program
enrollment vary by level of undergraduate educational
debt?
 Contributions to Current Body of Research
 Utilize propensity score methods to control for self-
selection bias
 Control for differing effects between types of financial
support
 National dataset includes students from all master’s
degree program areas
 Dataset: National Postsecondary Student Aid Study 2008
What’s Next?
 January – Submission for an AIR Research Grant
 January to May – Conduct analyses
 May – Submission for research presentation at ASHE
 December – Submit to journal for publication