documents oerc_160913_va_symp_thorn
TRANSCRIPT
Synthesizing Extant Knowledge
for Practitioners in a Carnegie Knowledge Network
Chris Thorn, Managing DirectorAnalytics and Program Technology
September 16, 2013⦁ Columbus, OH
Triple Aims of Educational Improvement
2
EFFICIENCY
EFFECTIVENESS
ENGAGEMENT
Context: We Live in Extraordinary Times
More EfficientSystems
Ambitious Learning For All Students
MoreRelevance
Why focus on value added?
Value-added methods are relatively new, use is increasingly wide spread, but many technical questions remain
unresolved.
The Problem We’re Trying to Address:
• The state of knowledge in the field is changing rapidly
• The vast amount of information can be overwhelming
• Most findings are written in highly technical language
• Many experts are tied to commercial interests or policy stances
What a teacher interested
in learning more about
value-added might find
through an online search.
McCaffrey, D. F., Lockwood, J. R., Koretz, D.,
Louis, T. A., & Hamilton, L. (2004). Models for
value-added modeling of teacher effects. Journal
of educational and behavioral statistics,29(1), 67-
101.
Instrument Design Actual Practices of
Use
Rules & Regulations
Economists
Applied
Researchers
Designers
Statisticians
Policy
Advocates
LegislatorsState
Education
Officials
Union Leaders
Teachers
Principals
Local Teacher
Union Officials
District
leadersExternal Service
Providers
Carnegie’s Distinctive Role: Integrative Agent
The Carnegie Knowledge Networkwww.carnegieknowledgenetwork.org
• Identifies high priority areas characterized by significant knowledge gaps between research and practice
• Builds on an R&D agenda focused on practitioner needs
• Engages the community of practitioners
• Assembles balanced technical expertise
• Acts as an integrative agent
• Builds scholarly consensus
• Informs policy
CKN Online
Most common value added models in use
Vendor Name of Model Brief Description
American Institutes for Research (AIR)
VariedUsually control for student
background
Mathematica VariedUsually control for student
background
National Center for the Improvement of Educational
Assessment (NCIEA)
Student Growth Percentile (SGP) Models
Models a descriptive measure of student growth within a
teacher’s classroom
SAS EVAASModels control for prior test scores but not other student
background variables
Value Added Research Center (VARC)
VariedUsually control for student
background
Highlights of the recommendations
• Teachers of advantaged students benefit from models that do not control for student background factors, while teachers of disadvantaged students benefit from models that do control for student background factors
• Even when correlations between models are high, different models will categorize many teachers differently
• Rules for combining measures should reflect the qualities of those measures
Highlights of the recommendations
• High quality linkage is critical (dosage/teams/mobility)
• Consider the level of precision and balance the risks
• Bias may arise when comparing the value-added scores of teachers who work in different schools
• The properties of value-added measures differ across grades and subjects
• There is only a moderate, and often weak,
correlation between value-added calculations for the same teacher based on different tests
What’s on the Horizon for Carnegie
• We have little research to draw upon for designing systems or for predicting the effects of emerging
evaluation systems
• The Foundation leveraging the pressure of
accountability as the gateway drug to improvement
• Variation in effectiveness is the problem to solve
An Interesting Case Example
• First year results from a large randomized field trial
of Reading Recovery
(I3 initiative)
• Key: a multi-site trial
12
0
2
4
6
8
10
12
14
16
-0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9
Effect Size
RCT (average) Treatment Effect: Reading RecoveryN=141 schools
It is a success
0
2
4
6
8
10
12
14
16
-0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9
Co
un
t
Effect Size
Distribution of RCT Treatment Effects: Reading RecoveryN=141 schools
Undesirable/Weak Outcomes
Positive Deviants
0
200
400
600
800
1000
1200
F D C B A
Distribution of Letter grade of
Overall Value-Added for Ohio Schools
See the System to Improve it
We cannot improve outcomes without understanding the processes that generate them and the interconnections
between the processes.