Brian Zuckerman American Evaluation Association November 2 nd 2011
description
Transcript of Brian Zuckerman American Evaluation Association November 2 nd 2011
It's an Evolution: Changing Roles and Approaches in the Evaluation of the
Pittsburgh Science of Learning Center
Brian Zuckerman American Evaluation Association
November 2nd 2011
Pittsburgh Science of Learning Center
• Purpose: Leverage cognitive theory and computational modeling to identify the conditions that cause robust student learning.
• Goals: Fundamentally transform– translational research in education– generation of learning science theory
Ed technology + Wide dissemination = “Basic research at scale”
+ =
PI: Ken Koedinger (CMU)Co-PIs: Chuck Perfettti (UPitt), David Klahr (CMU), Lauren Resnick (Upitt)
PSLC: Transforming Translational Research
• LearnLab = social & technical infrastructure to support field-based basic research– Controlled experiments in real courses– Educational technologies => Data!
• Practice-relevant discovery– What lab-based theory survives
translation?– Field-based data drives discovery
English Reading Tutor
Chemistry Virtual Lab
Researchers
LearnLab
Schools
Algebra Intelligent Tutor
PSLC: Transforming Theory GenerationEmerging “Computational Learning Science”• Data mining & computational modeling techniques • New data sources: Brain imaging, classroom video,
student interactions with ed tech
pslcdatashop.web.cmu.edu/KDDCup/
PSLC Capacity Building• Vast student data repository• New field: Educational Data Mining • 2010 KDD Cup
– Annual competition of Knowledge Discovery and Data Mining conference
– Task: Predict step-by-step performance of 10,000 algebra students across school year
PSLC: Research to Practice
• Translation is built in– LearnLab embeds experiments &
scientific data collection within running courses
– Many outcomes of 200+ learning studies incorporated into courses
• Ed tech dissemination partners– Carnegie Learning, Inc.
>600,000 K12 math students a year– Open Learning Initiative
1000s of college student users a semester
Cognitive Tutor 2010/11 release uses PSLC results (Butcher & Aleven, 2008)
In use in all 50 states
PSLC Activities Relative to Program Goals
• Conducts large-scale research– Large-scale theory development– Formation/nucleation of new fields or subdisciplines
• Develops and maintains infrastructure useful to community– Large-scale infrastructure development (LearnLab, DataShop, tools)
• Educates diverse, highly competent, and globally-engaged workforce– Education of graduate students and postdoctoral researchers– Broadening participation (e.g., PSLC summer internships)– Conducts center mass-requiring ancillary education efforts for broader
learning sciences community (e.g., PSLC summer school)• Forges valuable partnerships
– Among PSLC researchers in interdisciplinary collaborations– With industry/external stakeholders
7
Changes in PSLC Organization
• Reorganization around renewal– Four clusters become three thrusts– Change in co-PI, on Executive Committee
8
Evaluation Context
• Evaluator context– STPI funded by the center– STPI came on board around first site visit in 2005– Center passed through five-year review, now in year 7
• Shifts in PSLC activities and logic model– Change in organization– Shifting emphasis on goal of theoretical framework
development– Other smaller changes (e.g., shift in diversity goals
toward long-term expansion of field)
Evolution of Evaluation Effort Matches Changes in PSLC Lifecycle
• Years 1-2: Predominantly focused on growth of “Centerness” with data collection internal to Center– Management processes (interviews)– Collaboration formation (interviews, collaboration survey)– Development of Center-wide language and culture (interviews)
• Years 3-4: In preparation for site review focus shifted– External investigators’ knowledge of Center research and predictions of future value– Theoretical framework development/wiki analysis– Bibliometric analysis of publications to date
• Years 5-6: Center reorganization led to refocus on “centerness”– Return to interviews in Years 1-2– Analysis of changes in thrust plans over time
• Present: Evaluation effort largely dormant until plans for SLC-wide evaluation become evident– Some continuing activities around sustainability
Data Collection Changes Over Time (Partial list of measures and approaches)
Topic Data Collection Strategy Years 1-2
Years 3-4
Years 5-6
Theoretical framework Internal Interviews/growth of common language
X X
Theoretical framework Analysis of PSLC theory wiki X
Theoretical framework External interviews/value of PSLC theory development to date
X
Theoretical framework Analysis of thrust plans X
Theoretical framework Bibliometrics X X
New fields External interviews/role of PSLC in educational data mining community
X
Value of infrastructure Internal interviews/use internally of PSLC DataShop, tools
X X
Value of infrastructure External interviews/knowledge of PSLC DataShop, tools
X
Data Collection Changes Over Time (cont.)
Topic Data Collection Strategy Years 1-2
Years 3-4
Years 5-6
Education of students Internal interviews/perception of value of PSLC participation
X X
Education of students Tracking next steps of PSLC graduates
X X X
Broadening participation Participant observation and interviews with interns
X X
Broadening participation Tracking next steps of interns X X
Other ancillary educational efforts
Follow-up up with participants in summer school
X
Data Collection Changes Over Time (cont.)
Topic Data Collection Strategy Years 1-2
Years 3-4
Years 5-6
Collaboration formation Internal interviews/value and success of center-wide collaboration formation approaches
X X
Collaboration formation Internal Interviews/ collaborativeness in Center
X X
Collaboration formation Collaboration survey X
Collaboration formation Bibliometrics X X
External collaborations External interviews/value of collaboration
X
Management Value and effectiveness of center structures and processes
X X
Reflections
• Evaluation in complex context– Pressure from government funder to demonstrate
results even during first five-year period– Change in PSLC organization and goals over time
• Required nimble evaluation approach as a result– Evaluation plans developed in Years 1 and 5 served
as point of departure rather than blueprint– Difficult-to-maintain balance between need for
continuity in data collection and shifting priorities