1 Data-Driven Instruction: A Case for Adaptive LearningNational Academy of Engineering - Frontiers of Engineering Education Irvine, CA - September 27, 2016 Mary Besterfield-Sacre Nicholas A. DeCecco Professor Department of Industrial Engineering Director, Engineering Education Research Center Swanson school of Engineering Center associate – Learning research and development center University of Pittsburgh

2 Lots of Data Personalized

3 Delivery and Learning CycleCurrent Pre-work Reading Direct Teaching In-class lecture Engaged Learning without Instructor Homework, project, study… Flip Model Pre-work Quiz Pre-HW Reading Direct Teaching Video Engaged Learning with Instructor Continued Engaged Learning without Instructor Homework, project, study…

4 Flipping – perspective of Data DrivenModel Pre-work Quiz Pre-HW Reading Direct Teaching Video Engaged Learning with Instructor Continued Engaged Learning without Instructor Homework, project, study… Flipping – perspective of Data Driven How does the data change how you teach?

5 Improving and Assessing Student Learning in an Inverted STEM Classroom Setting, NSF/DUE–1322586Autar Kaw – USF Mary Besterfield-Sacre – PITT Renee Clark –Pitt Yinyang Lou – ASU Andrew Scott – AAMU

6 Theoretical frameworkPERFORMANCE theory of learning for the digital age Connectivism (Siemens, 2005) ENJOYMENT social, interactive, and cooperative nature of learning DIVERSE LEARNERS Social Development (Vygotsky, 1978) Cognitive Apprenticeship (Collins et al, 1989) students learn through the guidance, support, or help of an expert WHO - WHAT HOW OUTCOME

7 The experiment

8 Theoretical framework- Blended PERFORMANCE Connectivism (Siemens, 2005) Pre-req matl 24/7 open courseware Piazza discussion board Clicker quizzes Peer-to-peer discussion Re-poll Active learning Short exercises w/ peers 2-way questioning Lecture Automatically graded quizzes Problem sets 2-3 programming projects In class exercises ~ HW DIVERSE LEARNERS Social Development (Vygotsky, 1978) ENJOYMENT Cognitive Apprenticeship (Collins et al, 1989) CUCEI Survey Focus Grp WHO - WHAT HOW OUTCOME

9 Theoretical framework- Flipped PERFORMANCE Connectivism (Siemens, 2005) Video lectures 24/7 open courseware Piazza discussion board Graded quiz Essay question Clicker quizzes Peer-to-peer discussion Re-poll Active learning Exercises & Outline solution w/ peers Micro-Lecture based on pre-quiz & essay Automatically graded quizzes Problem sets 2-3 programming projects In class exercises ~ HW DIVERSE LEARNERS Social Development (Vygotsky, 1978) ENJOYMENT Cognitive Apprenticeship (Collins et al, 1989) CUCEI Survey Focus Grp WHO - WHAT HOW OUTCOME

10 Across Multiple Demographic Groups –All Schools Flip Blended Sample Size All students 180 215 Female 45 37 CC Transfer w/ Assoc. 42 60 URM 71 54 Pell Grant recipient 63 70 Are there differences in performance level for various demographic groups? Across Multiple Demographic Groups – Not statistically significant; small effect sizes Individual Schools – Lower Bloom’s USF or ASU - Flipped instruction was slightly better for multiple- choice performance AAMU – Large differences in favor of the blended approach Mixed results from other studies

11 What’s missing

12 A case for adaptive learningNational Academy of Engineers 14 Grand Challenges Advanced Personalized Learning Bloom (1984) showed tutoring makes the average student perform above 98% of students in conventional class (Cohen d=2) $$$$ Adaptive platforms Knewton Adaptware Smart Sparrow Combine Internet, optimization algorithms, machine learning, content graphing, statistics, and learning sciences

13 Each student comes to class from a different pointWe want to “equalize” that

14 Theoretical frameworkKnewton Connectivism (Siemens, 2005) Prior knowledge  Organization of knowledge  DIVERSE LEARNERS Social Development (Vygotsky, 1978) Mastery Practice with feedback   Cognitive Apprenticeship (Collins et al, 1989) Transforming a Flipped STEM Course Through Adaptive Learning, NSF WHO - WHAT HOW OUTCOME