SPARC SelfPaced Learning Increases Retention and Capacity Teaching

SPARC: Self-Paced Learning Increases Retention and Capacity Teaching fairly and teaching well Jeff Offutt Software Engineering / Computer Science https: //www. cs. gmu. edu/~offutt/ Supported by Google’s 3 X Capacity Program

Would you believe me if I told you 2 that reading powerpoint aloud is not the same as teaching © Jeff Offutt Supported by Google’s 3 X Capacity Program

Who Am I? • Country boy from Appalachia • Dad: Steffi (26), Joyce (22), Andrew (14) • Husband: Jian • Instructor: College at all levels • Coach: Basketball, Odyssey of the Mind, Baseball r e h c a Scientist: >175 published papers e T • Editor: Research journal • • Author: Software testing book 3 © Jeff Offutt Supported by Google’s 3 X Capacity Program

George Mason University • 34, 904 students (Fall 2016) – – 24, 000 undergraduate 8, 200 MS 2, 200 Ph. D 500 Professional • Most diverse university in the country • 148 buildings – plus 4 under construction – Campuses in Prince William County, Arlington, and Korea • Opened as branch of UVA in 1957 – Full university in 1972 – About 12, 000 students in 1992, 20, 000 in 1999 • Total operating budget : $1, 001 million – Research expenditures : $102 million 4 © Jeff Offutt Supported by Google’s 3 X Capacity Program

My Background Tools of the trade chalk whiteboar ds ppt emai l My evolution Demand er 5 Judg er Collaborat or © Jeff Offutt intern et l i t n u Up 3 1 0 2 ~ Leader Encourag er Supported by Google’s 3 X Capacity Program

Eric Mazur Taught Me Better Students learn only slightly more from great lecturers than from poor lecturers Students learn a lot more from active learning exercises 6 © Jeff Offutt Supported by Google’s 3 X Capacity Program

Giving lectures is not teaching But they are addictive And slides become a crutch h c ea t ; s s e l e k l r o Ta m 7 © Jeff Offutt Supported by Google’s 3 X Capacity Program

5 -Min. Small-Group Discussion Do you believe this? Do you like to lecture? Can you say “cognitive dissonance”? 8 © Jeff Offutt Supported by Google’s 3 X Capacity Program

Benefits of Active Exercises • More student engagement – Students stay awake—passively listening is hard ! • Increased collaboration – An important and useful skill post-college – Many students enjoy learning together • Increased peer-learning – Students can learn a lot from each other – They can help each other work through difficult concepts and can often understand their difficulties better than professors do • Increased attendance – The return-on-investment analysis becomes much easier • Students feel they are respected 9 © Jeff Offutt Supported by Google’s 3 X Capacity Program

Types of Active Exercises Discussion Writing Examples Problemsolving Games or role playing 10 © Jeff Offutt Supported by Google’s 3 X Capacity Program

But … Switching is Hard! “Ya can’t teach an old dog new tricks. ” Habits die hard We have huge investments in lectures Lots of powerpoint slides How do we manage this kind of change? 11 © Jeff Offutt Supported by Google’s 3 X Capacity Program

Thus, My Current Motto Talk less; teach more I’m trying to be self-persuasive … 12 © Jeff Offutt Supported by Google’s 3 X Capacity Program

Challenges • Timing the class is harder with active exercises • The classroom infrastructure is important – Tables, movable desks, immovable desks, rising rows • Encouraging collaboration can be challenging – Many students are not used to collaboration – Some do not play well with others – Some students prefer competition • Encouraging diversity of thinking – A start of an argument or an opportunity for growth? • Changing your habits – Preparation and your experience in the classroom is very different • Your interaction—deciding when to step in 13 © Jeff Offutt Supported by Google’s 3 X Capacity Program

Classroom Layout Matters 14 © Jeff Offutt Supported by Google’s 3 X Capacity Program

Full-Room Discussion How can you do this in your classes? Questions about in-class exercises? How much time do you spend on in-class exercises? How often did your college professors use in-class exercises? 15 © Jeff Offutt Supported by Google’s 3 X Capacity Program

My Next Step in Evolution Why is learning to program hard? Logical thinking ? Poor tools? NO! It’s too complicate d? We punish students for the most natural way to learn: Collaboration 16 © Jeff Offutt Supported by Google’s 3 X Capacity Program

Cheating I think most people are honest But many will cheat if they think the system is rigged against them 17 © Jeff Offutt Supported by Google’s 3 X Capacity Program

Small-Group Discussion How much cheating do you find in your classes? 18 Are they truly dishonest people? © Jeff Offutt Supported by Google’s 3 X Capacity Program

SPARC Self-PAced Learning increases Retention and Capacity George Mason University is replacing the “traditional” course model for CS-1 and CS-2 with an innovative “black-belt” teaching model that features: ● Self-paced learning ● Peer instruction ● Separate collaborative practice and individualized assessments ● Flipped classrooms ● Automated grading (JUnit) ● Multiple attempts at assessments 19 © Jeff Offutt Supported by Google’s 3 X Capacity Program

SPARC Team Paul Ammann, Kinga Dobolyi, Chris Kauffman, Jaime Lester, Jeff Offutt, Upsorn Praphamontripong, Huzefa Rangwala, Sanjeev Setia, Pearl Wang, Liz White 20 © Jeff Offutt Supported by Google’s 3 X Capacity Program

Challenges from the Surge • Retention of students • Retaining educational quality • Our definition of “traditional courses” – Use the same assignments for practice and assessment • Collaboration is always cheating • Catching cheaters a major priority – All students must complete assignments synchronously – Class meetings are almost entirely lectures • First year classes are diverse – Gender, ancestry, pre-knowledge, learning styles & speed • Fourth year classes are not diverse 21 © Jeff Offutt Supported by Google’s 3 X Capacity Program

Impacts on Diversity • Increasing enrollment often leads to competition – Classes, TA time, professor time, grades, … • Educational research indicates average differences – 17 -22 year old white and asian males thrive under competition – 17 -22 year old females and minorities thrive with collaboration • Studies from NCWIT, SACS, and NSF conclude that “just say no” introduces gender- and racial-bias • Ways to encourage diversity : – Collaboration, not competition – Peer learning – Encouraging success 22 – Building resilience – Obviating cheating – Self-pacing – Multiple attempts © Jeff Offutt Supported by Google’s 3 X Capacity Program

The SPARC Model • Self-paced demonstration of skills and knowledge • Online content delivery (flipped) • Ten stages (belts) in a course – Each stage has multiple practice and assessment problems • Students practice programming during class – Collaboratively (peer learning) – Instructors and TAs offer help and “mini-lectures” • Students demonstrate skills individually in labs (assessments) – Self-paced learning – Grading is automated (Junit) – Up to five attempts are allowed • Automated intervention alert system 23 © Jeff Offutt Supported by Google’s 3 X Capacity Program

Full-Room Discussion What do you want clarified? Have you used any of these techniques? 24 © Jeff Offutt Supported by Google’s 3 X Capacity Program

SPARC Classroom 25 © Jeff Offutt Supported by Google’s 3 X Capacity Program

SPARC Resources SPARC website UTA training materials (classroom & labs) SPARC program problems (practice & assessment) PCE training materials The Ma. GE project (Mount Holyoke) 26 © Jeff Offutt Supported by Google’s 3 X Capacity Program

What Is Working Really Well • Collaboration • Peer learning (UTAs) • Having multiple attempts to pass • Separating assessment from practice • No cheating • Multiple attempts • Self-pacing (with caveats …) 27 © Jeff Offutt Supported by Google’s 3 X Capacity Program

What Has Been Hard • Breaking the semester barrier – Incompletes, in-progress, course hours • Procrastination versus a legitimate need for more time to master programming • Software is hard to build! – Especially at universities • Recruiting good UTAs • Comparing SPARC sections with non-SPARC sections – Gathering data • Getting the right learning spaces 28 © Jeff Offutt Supported by Google’s 3 X Capacity Program

Data and Observations Numeric comparisons with “traditional” classes CS 1, 8 sections, 4 semesters > 550 students Cheating reduced from 10%-20% to one Pass rate slightly higher by the end of semesters (same instructor) Many students pass after end of semesters Similar scores on shared final exam questions 29 © Jeff Offutt Supported by Google’s 3 X Capacity Program

Data and Observations • Survey data – Final grades significantly correlated with : • Perceived helpfulness of UTAs • Perceived helpfulness of sample assessments • Perceived helpfulness of practice problems • Anecdotes – SPARC students say the class is much less stressful – SPARC students feel they are respected – Teacher finds this method much more satisfying – Student-teacher relationship is much better – Instructor doubled the number of sections with only a minor increase in workload 30 © Jeff Offutt Supported by Google’s 3 X Capacity Program

Final Discussion What pieces do you think might work for you? What do you think won’t work? 31 © Jeff Offutt Supported by Google’s 3 X Capacity Program

Contact Jeff Offutt offutt@gmu. edu http: //cs. gmu. edu/~off utt/ Talk less; teach more 32 © Jeff Offutt Supported by Google’s 3 X Capacity Program