How Much of a “ Running Start ” Do Dual Enrollment Programs Provide Students?

How Much of a “Running Start” Do Dual Enrollment Programs

Provide Students?

James Cowan & Dan GoldhaberCenter for Education Data & Research (WWW.CEDR.US)University of Washington Bothell

Background on Dual Credit Programs

• Dual-credit programs, which allow students to earn college credits while still enrolled in high school

• Rapidly become a popular college preparatory program• 1.2 million nationally • Seen by many as a cost-effective way of

increasing college readiness and college enrollment among high school students

• Very little empirical research on whether these programs effect college-going behavior

2

Prior Research on Dual Enrollment Programs• Most studies compare the outcomes of dual enrollment

students to other high school students using regression techniques to control for a variety of covariates (Allen & Dadgar, 2012; An, 2012; Karp et al., 2007; Swanson, 2008) and find large positive effects (12-17% pts) on college enrollment• Many use controls for academic proficiency that may be

endogenous• Speroni (2011) exploits a GPA eligibility requirement for

Florida’s dual enrollment in a regression discontinuity design and finds no statistically significant effect of dual enrollment on college attendance or completion

3

Three Mechanisms Through Which Dual Enrollment May Affect Students

The thickness and shading of the arrows indicates the propensity of students to go to college

Two-year college Four-year college

Least proficient Average proficiency Most proficient

COLLEGE READINESS DISTRIBUTION

4

First Mechanism: Cost

Two-year college Four-year college

COLLEGE READINESS DISTRIBUTION

Least proficient Average proficiency Most proficient

Cost to students of college credits reduced

5

Second Mechanism: Information

Two-year college Four-year college

COLLEGE READINESS DISTRIBUTION

Least proficient Average proficiency Most proficient

Students learn about ability to do college work and their fit with two- and four-year college

6

Third Mechanism: Socialization and Investment Accumulation

Two-year college Four-year college

COLLEGE READINESS DISTRIBUTION

Least proficient Average proficiency Most proficient

Peer influence and banking of degree

7

The Running Start Program in Washington

• Allows juniors and seniors in public high schools to take courses at all community colleges tuition-free• Tuition is paid by school districts, which pays the

receiving community college about $4500/FTE in 2010 (93% of the state basic education allotment• Estimated to be about 60% of the cost of course provision• The state estimates the total tuition subsidy cost $41.3 million for

the 2009-2010 school year (about $2200/RS student subsidy)

• School districts not permitted to put conditions on Running Start participation, but colleges determine eligibility based on Accuplacer or high school GPA

8

Investigating the Effects of Running Start• We focus on five outcomes:

1. Probability of high school graduation2. Probability of college enrollment3. College major4. College persistence5. College credit accumulation

• Try to identify causal effects of Running Start:1. OLS models that adjust for observable covariates2. IV estimates that rely on variation in geographic

proximity to RS provider + specific course offering3. RD based on GPA cutoff

9

Running Start Data

• Over 200,000 (11th and 12th grade) students starting over four school years, 2005-2006 through 2008-2009• 12% of HS students over this period take at least 1 RS

course• Data limitations: enrollment in a post-secondary

institution is limited to the 34 community and technical colleges, four of the five in-state public universities; no data on enrollments in private colleges or out-of-state public universities• But, the above sample covers nearly 75% of students

going on to college in Washington State

10

Just in Case…

• OLS models suggest significant Running Start impacts on most outcomes• Large estimated effects on college enrollment,

major, credit accumulation• Little estimated effect on college persistence• Negative effect on standard high school graduation

• Above findings are quite sensitive to sample and conditioning variables

• IV and RD results are consistent with above findings, but very imprecisely estimated so we cannot rule out null of no RS effect

11

Selected Sample Means: Outcomes 1 Year After High School Graduation

RS Students Non-RS Students All

Outcomes 1 Year After High School Graduation: Full-time college 0.43 0.25 0.27

(0.49) (0.43) (0.44)

Full-time two-year college 0.23 0.11 0.12

(0.42) (0.31) (0.33)

Full-time four-year college 0.21 0.14 0.15

(0.41) (0.35) (0.36)

Regular HS Diploma 0.89 0.8 0.81

(0.31) (0.40) (0.39)

Adult HS Diploma 0.01 0.01 0.01

(0.11) (0.09) (0.10)

GED 0.01 0.02 0.02

(0.07) (0.13) (0.12)

Full-time four-year college 0.23 0.13 0.14

(0.42) (0.33) (0.35)

12

Selected Sample Means: Outcomes 3 Years After High School GraduationOutcomes 3 Years After High School Graduation:

Full-time four-year college 0.22 0.01 0.04

(0.42) (0.11) (0.19)

Post-secondary credits 71.02 22.47 28.48

(46.14) (34.00) (39.14)

STEM major 0.04 0.01 0.02

(0.20) (0.12) (0.13)

Undeclared major 0.01 0.01 0.01

(0.07) (0.10) (0.09)

13

Student Controls

Student Controls: Student race: Asian 0.1 0.08 0.08 (0.31) (0.27) (0.28) Student race: Black 0.03 0.05 0.04 (0.17) (0.21) (0.20) Student race: Hispanic 0.05 0.11 0.1 (0.21) (0.31) (0.30) Student has disability 0.01 0.11 0.1 (0.12) (0.31) (0.30) Student gender: Female 0.6 0.48 0.49 (0.49) (0.50) (0.50) GPA (10th grade) 3.17 2.66 2.73 (0.65) (0.88) (0.87) FRL Status (10th grade) 0.19 0.31 0.3 (0.40) (0.46) (0.46) Bilingual Status (10th grade) 0.01 0.05 0.04 (0.11) (0.21) (0.20) 10th grade reading WASL 0.58 0.03 0.09 (0.75) (0.95) (0.94) 10th grade math WASL 0.62 0.04 0.11 (0.75) (0.96) (0.96) 10th grade science WASL 0.53 0 0.06 (0.73) (0.93) (0.92) 10th grade writing WASL 0.53 0.03 0.09 (0.63) (0.94) (0.92) Running Start participant 1 0 0.12 (0.00) (0.00) (0.33) Observations 24788 175527 200315

14

OLS Point Estimates: Different Types of HS Graduation

15

OLS Point Estimates: College Enrollment

16

OLS Point Estimates: College Major

17

OLS Point Estimates: Persistence and Accumulation

18

Heterogeneous Effects of RS19

Heterogeneous Effects of RS20

Heterogeneous Effects of RS21

IV AND RD Results22

• Instrument with measures of access to Running Start• Program w/in 2 or 3 miles; Online US History Course

• Use GPA eligibility requirement at 2 colleges for RD

• RD & IV results very imprecise• Suggest larger negative effect on enrollment in 4-

year college initially and after 3 years• In most cases, results not inconsistent with

enrollment effects of the magnitudes observed in previous studies

Policy Implications

• Some samples and control variables are endogenous to the choice of RS participation and the use of these affects the estimated magnitudes of the RS effect

• Reluctant to draw strong conclusions given the divergence between OLS and both IV and RD results• More data coming soon (additional cohorts and

private in-state universities) which should help

23