280 likes | 350 Views
Are Students Dropping Out or Dragging Out the College Experience? & Modeling the Enrollment Path. Leslie S. Stratton & James N. Wetzel Virginia Commonwealth University Research funded in large part by an AIR Research Grant 2012 Presentation at NEAIR Conference. Plan. Motivation
E N D
Are Students Dropping Out or Dragging Out the College Experience? &Modeling the Enrollment Path Leslie S. Stratton & James N. Wetzel Virginia Commonwealth University Research funded in large part by an AIR Research Grant 2012 Presentation at NEAIR Conference
Plan • Motivation • Briefly review some literature • Data • Goals of each paper • Results of Paper #1 • Results of Paper #2 • Conclusion
Motivation • Completion rates are a concern. • Enrollment is up, but graduation rate is unchanged (63%). • 45% for African Americans, 54% for Hispanics, 54% for low income, and 53% for first generation college students. • Benefits accrue primarily upon graduation. • Costs are incurred upon enrollment. • Shared by students/families and public. • Dept of Education is targeting schools with low graduation rates.
Literature • Studies of college outcomes are proliferating. • Many control for socioeconomic status: • Paulsen, St. John (2002), Swail, Cabrera, Lee (2004), Ishitani (2003, 2006), Titus (2006), … review by Kuh et al. (2006) & find these populations are disadvantaged. • Controlling for academic background is even more important • Adelman 2004, Carneiro & Heckman 2002 find background trumps race/ethnicity.
Literature • But, many studies are • based on data from 1 university, • do not adequately control for academic background, or • designate success as graduation within a particular time horizon. • Not all those who have not graduated in X years have given up. • Many students enroll PT, stop out, transfer.
Contributions of 1st Paper • Multi-institution study. • Controls for SES & academic background/ability. • Distinguishes between those not enrolled and those still enrolled at 6 year mark. 2nd Paper • Also controls for enrollment path.
Data • 1996/2001 US Beginning Postsecondary Survey • National sample of those beginning in 1995-96 academic year. • Restrict sample to those: • Who are followed through 2001 • Who initially attend 4 year institutions • Who are age 23- & from US • Sample of about 5820 students.
Research Questions: Paper #1 • Most studies compare graduates with non-graduates, BUT • 1) How common is persistence at the 6 year mark? • 2) How does distinguishing between those still enrolled and those not enrolled alter results of college outcome models?
Covariates I • Demographics: • Gender • Race/Ethnicity • Age • Marital status • Parental status • Parents’ education • Household income
Covariates II • Academic background/ability: • Standardized test scores • High school GPA • High school program of study • Unemployment rate. All observed at time of matriculation.
Question 1: How common is persistence at the 6 year mark? • 63.2% have graduated. • 23.4% are not enrolled. • 13.4% are still enrolled. • 36% of those who have not graduated are still enrolled in last term observed! • Persistence is common!
Question 2: How does controlling for persistence alter outcome models? • Standard analyses use logit specification to distinguish between graduates and non-graduates. • We use a MNL model to distinguish among 3 outcomes: graduation, persistence, drop out.
Distinguishing between Persistence and Non-Enrollment is Important • Statistically persistence is a distinct state. • Results by SES: • Disadvantaged students are less likely to graduate. • Hispanics are persisting. • 1st Generation college students are dropping out. • African Americans and Women are split. • Lower income students have mixed results.
Distinguishing between Persistence and Non-Enrollment is Important • Biggest Factors = Academic Background • Those with less than stellar high school GPAs are substantially less likely to have graduated. • Between 15 and 25% of these non-graduates are still enrolled. • Math background and SAT scores have a smaller marginal effect on graduation, but a similar association with persistence.
Evidence re-Graduation • Analysis of similar sample from Baccalaureate and Beyond survey suggests half of those still enrolled may graduate within a few years. • Analyses of college outcomes should look at persistence! • Is ‘dragging out’ worthwhile?
Why the Differences? • Enrollment paths are different. • Hispanics are more likely to enroll part-time. • Income/household constraints may favor part-time or stop out behavior. • Less prepared students may take more remedial classes, fewer courses/term. • Can only address this by modeling path as well as outcome 2nd Paper.
What Path Looks Like • Initial Enrollment • ~ 4% Part-time, 96% Full-time. • 1 Year Mark • ~ 11% Not Enrolled, 4% Part-time, 85% Full-time. • 3 Year Mark • ~19% NT, 7% PT, 71% FT, and 2% Graduated. • 5 Year Mark • ~ 22% NT, 5% PT, 13% FT, and 60% Graduated.
A Discrete Time Hazard of Enrollment • Takes initial FT/PT enrollment decision as given. • Models transitions from: • FT to FT, PT, NT, & graduation • PT to FT, PT, NT, & graduation • NT to FT, PT, & NT. • Graduation is treated as an absorbing state.
Specification • Uses a MNL specification to model these movements. • Models transitions from state j to state k as a function of • X = individual specific & matriculation characteristics, • Wt = time varying covariates, • Zt = past behavior, & • Θcaptures unobserved heterogeneity.
Specification • X includes ~ all covariates from paper 1 + info on first term attended (PT/FT, Spring/Fall, Semester/Quarter, …), • Wtincludes unemployment level and change, college grades, family chars, • Zt includes quadratic in time spent enrolled PT, FT, NT in past, & • Θ is iid normal.
Results: Predicted Path • Fixes Wt – College grades generally rise. • Initial Enrollment: Fixed. • 1 Year Mark • ~ 12% Not Enrolled, 5% Part-time, 83% Full-time. • 3 Year Mark • ~20% NT, 7% PT, 72% FT, and 2% Graduated. • 5 Year Mark • ~ 27% NT, 6% PT, 17% FT, and 50% Graduated. More NT & FT, Less Graduated
Results: Socio-economic indicators • Hispanics spend more time PT and NT than non-Hispanics. • 1st Gen spend more time PT and NT than non-1st Gen. • Blacks spend more time FT, but less time NT than whites. • Those from lower income HHs, spend more time NT. • Even controlling for path, all have a lower probability of graduating.
Results: Educational Background • Still much more important, especially high school grades. • Low performers spend less time FT, more NT and PT. • Controlling for path, effect on probability of graduation is modest.
Results: Unemployment Rate • Both level of unemployment and recent changes are significant. • Consistent with an opportunity cost argument: • A lower level of unemployment increases NT, while decreasing PT & FT. Little change in graduation rate. • Decreases in the unemployment rate appear to reduce PT enrollment and speed graduation.
General Conclusions • Persistence 6 years after matriculation is substantial and should be recognized and studied. - Is dragging out the college experience wise? • Enrollment paths constitute a complex problem. • Disadvantaged groups appear to have lower graduation rates, even controlling for path. • Academic background may play a greater role in driving the path rather than the outcome. • Economic conditions will change paths and outcomes.