School dropouts A framework for public policy analysis

School dropouts: A framework for public policy analysis Joshua Hawley, The Ohio State University

What do we mean by dropouts? • There are several useful definitions that can help… – 1) the broadest measure of dropouts are any student who has not finished a degree by the time the rest of the class has finished (four or five years) – 2) narrower version, any student who has not finished a degree at the school they started – 3) narrowest version, any student who does not finish some intermediate measure (first year of school)

Why should we care? • Students that quit school all together or never finish have used resources of the state, therefore if they leave your school has lost money/talent/future productivity. • Dropouts can be damaging to students, leading to attenuated college completion, low employment outcomes, and even suicide. • Dropouts are like the “canary in the coal mine” they tell us something is wrong with our product.

Other ways of thinking • In a sense there are two hypotheses about dropouts. • 1) College is a “cooling out” period for students, and drop out isn’t really bad for college. It just helps student clarify “choice” of career. • 2) College is an important social and econ. Investment, and like any investment, you want “high returns” with “high quality”

My way of thinking • I am of course on the side of the view that this is an investment and that we should strive to maximize the positive benefits of these investments… • Moreover, I feel that looking at dropout rates enables departments/majors to focus on comparing best practices that lead to some gaining high outcomes. • In other words, we can learn from each other about institutional good performance.

What is the U. S. graduation rate? Sector Graduated in 4 Years Graduated in 5 or 6 Years Total in 6 Years Flagships (most selective) 65% 21% 86% Flagships (mid level of selectivity) Flagships (less selective) 52% 28% 80% 33% 66% State systems (more selective) State systems (less selective) 54% 23% 77% 26% 25% 51% This table comes from the recent book by William Bowen, et. Al, Completing College: Finishing College at America’s Public Universities (Princeton UP, 2009)

Modeling Dropouts • Typical model for predicting dropouts is based on two methods: – Simple percentages (comparing the numbers that have finished vs. those that started, at some given point in time) – Multivariate methods • Logistic regression models looking at the “likelihood of completing” • Event history models looking a the survival rates to graduation using longitudinal data. • Propensity score matching, using matched sample to compare performance.

Questions Asked • What was the retention and (conversely dropout) of students through higher education after six years? • How did the retention (dropout) differ by age at which students started school? • What differences were there in student achievement of intermediate measures (first year and second year)? • How did the achievement of intermediate measures vary by age at which students start school?

Details about Ohio’s higher education system • Ohio is a large U. S. State (7 th largest in terms of population) about 11. 5 million people overall • Ohio’s Higher Ed system is managed in Columbus at the Board of Regents. • There are 13 universities (with 38 add. Branches) • 24 community and technical colleges • About 72, 000 degrees/annually; about 500 K enrolled

Ohio’s University System Cleveland Columbus Cincinnati OSU is a Research One school, the flagship campus

Strategic Plan • Interesting directions for the state in higher education – Overall, Ohio is experiencing increases in higher education enrollments at all levels – But, our 4 year graduation rates are low (about 60% overall complete in 4 years) • Therefore, to improve we have at least 3 key issues – 1) need to improve “access” of students (i. e. more non traditional students in pipeline). From 472 K to 700 K/annually – 2) need to improve “efficiency” of the system, produce more students with less wastage. – 3) need to improve “equity” of disadvantaged students (poor, minority, immigrant) in our schools

Example Study • Using data on all individuals that entered Ohio community and technical college system (this is our 2 year college system) • 45, 000 entrants in Year 2002 -3 • Supplement data with test score information and labor market data

How do we measure these increases over time? • Use state level data we compare on key metrics that have been developed by a national group (Achieving the Dream). • Measure achievement over time • Include all students (including those that enrolled for just one course) but didn’t get a degree or credential

Findings: overall 44, 995 students started in fall of 2003 40% were over age 24 in 2002 Overall, by 2009, 20% had completed a AA or BA degree; while 4% got a certificate. There are strong differences in these outcomes due to 1) Age, 2) part time status, and 3) school of attendance. Final outcomes for schools vary from a low of 16% to a high of 40%. Often school size was a contributing factor, with bigger schools experiencing lower completion rates

Findings: year 1 outcomes There is wide variation in achievement of key benchmarks in the first year of enrollment. In this case we examine 3. 1. 2. 3. Enrolled in first and second term Passed 80% of attempted hours Completed 24 + credit hours by spring of the first year • 60% of people enrolled in both fall and spring of first year (we already lost 40% of the total the first year!!) Only 24% of students completed at least 24 hours by end of first year About 2/3 passed 80% of more of the courses attempted • • For Ohio, enrolling fulltime makes all the differences

Findings: year 2 outcomes We have three year 2 outcomes 1. 2. 3. • • Enroll in the first and second fall of school Completed 24+ hours by end of year 2 Complete 48 + hours by end of year 2 About 50% of students enrolled in both the first and second fall they started college. Completion of 24 + hours and 48 + hours depends a lot on age at initial enrollment

Conclusions • Dropouts are a major issue for Ohio schools • Older students have particular issues with school dropout • Starting school as a full time student makes a difference • Final outcomes depend a lot on initial patterns of enrolling and completing courses.

Implications • There are several implications for Russian higher education institutions – 1) From a global perspective, universities (even selective ones like HSE) are paying attention to dropout rates. – 2) For a country with a declining demographic picture, drop outs can be a bigger issue as selectivity decreases when there is more competition for good quality students – 3) Measuring changes in drop out rates will require institutions level data – 4) Payoffs can be significant