sheeo policy conference august 9, 2012
DESCRIPTION
1. SHEEO Policy Conference August 9, 2012. Sandy Baum, David Wright, Lee Holcombe, Kim Hunter-Reed. @ HCMStrat. Project background. Started in 2011 by HCM Strategists, with support from the Bill & Melinda Gates Foundation - PowerPoint PPT PresentationTRANSCRIPT
1
1
SHEEO Policy ConferenceAugust 9, 2012
@HCMStrat
Sandy Baum, David Wright, Lee Holcombe, Kim Hunter-
Reed
2
Project background
Started in 2011 by HCM Strategists, with support from the Bill & Melinda Gates Foundation
Brought together researchers with different perspectives on higher education measurement
Asked research participants to think like policymakers and leaders
Asked policymakers and leaders to consider and respond to research
3
Key questions
How should we account for different student populations when evaluating higher education outcomes?
What methods are available or in use for policymakers and leaders to account for variations in inputs?What are the best ways to do it and what are the practical limitations?Can research findings be translated into policy?
4
Project components
Seven white papers Synopsis paper Literature review
Coming soon . . . Policy briefs Toolkit for states Outreach--sharing findings, including more voices, continuing to solicit input
4
5
Some general lessons and conclusions Input adjustments make a big difference Policymakers should make an effort to adjust for inputs when institutional comparisons have stakes attached
Imperfect input adjustments are better than none
Need to be clear about outcomes
6
Readers will find specific guidance on Important variables to include Examples of adjustment models and statistical techniques
Use of survey for input adjustments College learning assessments Strategies for working with messy administrative data
Strengths and weaknesses of U.S. News input adjustments
Many potential unintended consequences
7
Where to find out more . . .
Context for Success website:
www.hcmstrategists.com/contextforsuccess
Questions? Comments?
8
Advice for Moving Forward
Know the purpose, audience and questions to be answered
Consumers: Judging quality and value
Institutions: Improving
Accountability: Success and cost-effectiveness
9
Advice for Moving Forward
Comparing outcomes: Focus on similar institutions
Differences in credentials Goals of students Role of transfer
Selecting variables Preparation Demographics Attitudes, goals, behaviors
10
Context for Success White PapersCharles Clotfelter. "'Context for Success' Synopsis Paper."
Thomas Bailey and Di Xu. "Literature Review: Input Adjusted Graduation Rates and College Accountability."
Peter Riley Bahr. "Classifying Community Colleges Based on Students' Patterns of Usage."
Thomas Bailey. "Developing Input Adjusted Measures of Community College Performance."
Jesse M. Cunha and Darwin W. Miller. "Measuring Value-Added in Higher Education."
11
Context for Success White Papers (cont.)Robert Kelchen and Douglas N. Harris. "Can 'Value Added' Methods Improve the Measurement of College Performance?"
Stephen M. Porter. "Using Student Learning as a Measure of Quality in Higher Education."
John Pryor and Sylvia Hurtado. "Using CIRP Student Level Data to Study Input Adjusted Degree Attainment."
David Wright, Matthew Murray, Bill Fox, Celeste Carruthers, Grant Thrall. "College Participation, Persistence, Graduation, and Labor Market Outcomes."
An Input-Adjusted Framework for Assessing the Effectiveness of Tennessee’s
Higher Education Institutions
Context for Success project
Celeste Carruthers,a William Fox,a Matthew Murray,a Grant Thrall,b
David L Wrightc
a University of Tennessee, Knoxville; b University of Florida; c Tennessee Higher Education Commission
Background
• Policy relevance:
– Complete College TN Act ties HEI funding to student outcomes.
• Assumption:
– Funding incentives must take into account student characteristics that are likely to be important determinants of postsecondary outcomes.
SHEEO, 08-09-2012Context for Success paper
Research questions
• Purposes:
1) We statistically test whether some TN institutions were more effective than others at advancing the outcomes of 2002 entering first-time freshmen.
SHEEO, 08-09-2012Context for Success paper
Research questions
• Purposes:
2) We go on to test whether accounting for student inputs (ACT aptitude, age, low-income home neighborhood, LifeMode marketing profile) changes the ranking of institutions.
— Is model fit improved by supplementing or substituting our standard stock of administrative data with proprietary data on consumer tapestry profiles (called LifeModes) or with Census information?
SHEEO, 08-09-2012Context for Success paper
Method
• Cohort: 2002 first-time freshmen in TN public universities and community colleges
• Design:
– Multivariate OLS regression
– Used fixed effects to isolate institutional effectiveness
SHEEO, 08-09-2012Context for Success paper
Method• Compared models with four different sets of
predictor variables: 1) naïve model2) standard state agency administrative data3) Census neighborhood data4) proprietary LifeModes profiles
• Outcomes considered: student progression, transfer, graduation, and near-term labor market success
SHEEO, 08-09-2012Context for Success paper
Selected findings
• Timely completion:
– Females, higher entrance exam scores, and adult status were associated with more timely degree completion.
– Low-income status delayed completion.
– Additional semesters in college reduced earnings later on.
SHEEO, 08-09-2012Context for Success paper
Selected findings
• Earnings:
– Females and higher entrance exam score were associated with higher earnings in the years immediately following college.
– A significant black-white earnings gap favored white students.
SHEEO, 08-09-2012Context for Success paper
Six-year graduation rates (bachelor’s) across 4-year schools, relative to one omitted institution, unadjusted (blue) or adjusted for student and home characteristics
4-Year Institution
SHEEO, 08-09-2012Context for Success paper
Four-year graduation rates (associate’s) across 2-year schools, relative to one omitted institution, unadjusted (blue) or adjusted for student and home characteristics
2-Year Institution
SHEEO, 08-09-2012Context for Success paper
Observations and open questions• Institutions do matter: even after controlling
for all observed student inputs, some institutions are significantly more effective than others.– Also see Cunha and Miller in this series.
• Student inputs also matter: institutions’ apparent effectiveness changes when we account for student aptitude and background.
• The gains from more data are not always high.
SHEEO, 08-09-2012Context for Success paper
• Why are some institutions more effective than others? Institutional effects are a black box.
• Student goals and objectives may not align with traditionally tracked student outcomes.– Especially in community colleges!
• Assessing and comparing student growth across fields and campuses is challenging but not insurmountable.
SHEEO, 08-09-2012Context for Success paper
Observations and open questions
• Outcomes and institutional best practices should be included in performance-based funding. (p.6)
• Incentivizing colleges for student progression and completion relative to expectations is an improvement over unadjusted rewards…
• … but, be aware that outcomes will vary according to which students inputs are factored into expectations for colleges and universities. (p.32)
SHEEO, 08-09-2012Context for Success paper
Conclusions
A Project of the Texas Higher Education Coordinating Board
Input Adjusted Output Measures
Presented By: Lee Holcombe, Director, Higher Education Policy Institute
Analysis: Jesse Cunha, of the Naval Postgraduate School and Trey Miller of the RAND Corporation
26
Goals
Develop and implement a methodology that generates quantitative metrics of institutional performance.
Come as close as possible to causal estimates of an institution’s impact on a particular outcome.
Demonstrate benefits of collecting additional data.
Develop a set of concrete and actionable policy recommendations for practitioners and policymakers.
Higher Education Policy Institute
27
Outcomes
Persistence into 2nd year of college: Attain 30 credit hours at any public or private
four-year college in Texas by the end of the 6th academic year after graduating high school
Graduation: Receive a Bachelor’s degree at any public or
private four-year college in Texas by the end of the 6th academic year after graduating high school
Earnings: Sum of 4 quarterly earnings in the 8th calendar
year after high school graduation.
Higher Education Policy Institute
28
Data
Higher education Demographic variables, application and
acceptance data, credits attempted, degree completion
K12 (Secondary) Demographic variables, test scores, courses taken
Unemployment Insurance earnings
From College Board: SAT scores and survey info
Higher Education Policy Institute
29
Sample
Graduates of Texas public high schools in 5 cohorts (1998-2002)
Include those Texas public high school graduates who ever enrolled in a Texas public university (up to 2010).
217,723 enrollees, 169,239 of whom had earnings greater than $2,000
Higher Education Policy Institute
30
Findings
Large differences in unconditional outcomes across colleges diminish as each set of controls is added to the model.
Adding high school indicators and courses changes results much more than adding race and gender.
Different results across outcomes.
Measures are highly variable over time.
Higher Education Policy Institute
31
Policy Recommendations of Report
Assessment of institutional quality should involve a broad set of outcomes representing multiple dimensions of institutional performance.
Researchers should work with state policymakers to develop models that reflect state priorities.
Do not assume that results are causal.
Create broad classifications of institutional performance as opposed to explicit rankings.
Higher Education Policy Institute
32
Policy Response
Elusive quest for causality.
Excludes many non-traditional students.
Consider improvements within-institution
Dig inside the higher education black box.
Provide higher levels of information and support.
Use of more nimble data systems.
Consider learning outcomes.
Higher Education Policy Institute
33
Where to find out more . . .
COMING SOON
Context for Success website:
www.hcmstrategists.com/contextforsuccess
Questions? Comments?