Getting principled - Jeff Gray

GETTING PRINCIPLED
Reflections on Teaching CS Principles
at Two College Board University Pilots
SIGCSE
March 2017, Seattle WA
Jeff Gray, Michele Roberts, Jonathan Corley
Slides will be available at: http://bit.ly/sigcse2017-cspU
Overview
■ Motivation and Background
■ Pilots
– IUPUI
– UA
■ Pilot Results
■ Lessons Learned
Motivation
■ According to the College Board, CS Principles will
– “be the largest debut of an AP ever in the history of the
College Board, with nearly 50 percent more teachers
launching the program than the previous crown-holder:
the AP World History exam, which had 998 teachers in
its first year”
■ There is a growing request from high school CSP teachers for
universities/colleges to offer a mapping to the new AP CSP course,
with a targeted course number for AP credit
■ Our paper shares some of our experiences in creating such a course,
with a few lessons learned, that may be of interest to other CS
Departments considering a new similar CSP course in their curriculum
Attestation
http://www.csprinciples.org/home/about-the-project/attestation
The Need
Background – CSP Pilots
■ From 2010-2016 over various phases, the CSP course was piloted
■ Final Phase: 50 pilots (38 HS teachers and 12 university teachers)
■ Our paper focused on two of the university pilots:
University of Alabama and IUPUI
■ http://www.csprinciples.org/home/pilot-sites
2011-2012
2012-2013
2013-2016
Key Findings
■ College Board approved, and Haynie Research provided access, to the
student responses from our Pilot courses
– Our paper summarizes the past three years of our Pilot courses
■ Enrollment of women in the CSP course at each of our universities
was twice that of our Department’s traditional CS 1
– But one of our CSP courses showed less enrollment impact for
other URMs compared to the CS 1 course. Different from AP CSP
■ A significant difference was observed in the DFW
(Drop/Fail/Withdraw) rates for CSP compared to the CS 1 course.
– At both IUPUI and University of Alabama, the DFW rate for CSP is
only 2% and 3%, respectively, but for CS 1 it ranges from 18% to
54% at IUPUI and 50% to 61% at the University of Alabama.
■ Significant gains in pre-post CS self-confidence
IUPUI PILOT
Indiana University, Purdue University at
Indianapolis (IUPUI)
■ IUPUI is an urban research university located in downtown
Indianapolis
■ Downtown regional campus of IU’s main Bloomington campus and
Purdue’s main West Lafayette campus, with total student enrollment
at 30K, over 90% of which are state residents
Indiana University, Purdue University at
Indianapolis (IUPUI)
■ The CSP pilot was run by Senior Lecturer Michele Roberts from the
Computer Science Department (located in the School of Science).
■ The CS Department serves about 2,100 students a year (including
both majors and non-majors)
Course Launch
■
The CSP pilot launched in Spring 2014
–
Has been offered every Fall and Spring thereafter
■
The course launched with 22 enrolled students
■
The course is offered as a 3-credit, 200 level course for
non-majors
■
After a first successful semester, CSP was marketed to
advisors in face-to-face meetings using:
–
Syllabus
–
sample lab descriptions
–
student project displays
–
testimonials from students
The Course Growth and Evolution
■
In Fall 2015, school and university administration were successful in adding the CSP course in
the Analytical Reasoning category of state core educational requirements.
–
■
■
i.e., Approved to substitute for 3 of 9 required hours of mathematics
By Fall 2016, enrollment had grown
–
175 students (from 22 at launch)
–
5 full and waitlisted sections
In Fall 2016, CSCI faculty voted to count passing AP CSP scores of 3 or higher with CSP credit.
Pilot Administration
■ The IUPUI pilot was run with strong fidelity to the College Board
Curriculum Framework, delivering content within the CSP “Big Ideas”
■ Through Spring 2016 (when the pilot ended), CSP students submitted
the two Performance Tasks and completed a Final Exam
UA PILOT
University of Alabama (UA)
■ UA is a suburban public research university located in a college town
(Tuscaloosa), with an approximate total enrollment of 37,000
■ 46% of University of Alabama students are in-state residents
■ CS Department has approximately 700 undergrads/70 grads
CS 104
■ The CSP course at UA is offered as a 3-credit course within the
Department of Computer Science in the College of Engineering.
■ College Board CSP Pilot from 2011-2016 (UA and UW-Madison were
longest running university CSP Pilots)
■ Since 2013, the course is primarily enrolled by non-majors,
specifically Secondary Math Education (SME) and Math majors.
Pilot Administration
■ Programming Language: Snap!
■ Textbook: Blown to Bits and selected readings
■ Pedagogical approach: Cooperative learning structures
■ Similar to IUPUI, the UA pilot was run with strong fidelity to
the College Board CF. Performance Tasks done as HW; even
included former “Investigate” as an in-class activity
■ Spring 2016, UA formally recognizes CS 104 for AP CSP
mapping
■ 2016-2017 academic year moved to two offerings, with
waitlisted enrollment capped at 40
PILOT RESULTS
Broadening Participation (IUPUI)
Year
2013-2014
2014-2015
2015-2016
CSP
CS 1
CSP
CS 1 CSP
CS 1
Women
37%
13%
58%
12%
63%
22%
URM
8%
12%
18%
9%
19%
20%
■ Successful at Broadening Participation
– Female participation rose from 37% to 63% from first to most
recent offering
– Other URM (Latino and African Americans) rose from 8% - 19%
– These numbers compare more favorably with the more
homogeneous CS 1 population
Broadening Participation (UA)
Year
2013-2014
2014-2015
2015-2016
CSP
CS 1
CSP
CS 1 CSP
CS 1
Women
52%
27%
55%
26%
70%
28%
URM
30%
16%
21%
21%
13%
21%
■ Successful at Broadening Participation
– Female participation rose from 52% to 70% from first to most recent
offering
– Other URM DECREASED from 30% to 13%
– The gender numbers are much improved over CS 1, but URM interest
was disappointing
Student Success (DFW rates)
Year
2013-2014
2014-2015
2015-2016
CSP CS 1 CSP CS 1 CSP CS 1
IUPUI-DFW
2%
54%
2%
23%
2%
18%
UA-DFW
3%
52%
3%
60%
3%
51%
■ Pilots at both universities saw surprisingly low levels of DFW rates
(Drop, Fail, Withdraw rates)
– Less than 3%, which compares to CS 1 DWF rates of 50% or
higher
■ Pilots at both universities saw high levels of pass rates
– Pass rates were above 97%, 30% higher than pass rates for CS 1
Student Self-Confidence
CSP Students Experienced an Improvement in Confidence with CS!
■ Survey question: “I have confidence when it comes to CS”
■ Both CSP pilots observed a significant pre-post gain in student
confidence
IUPUI
UA
Student Interest in CS
CSP Students Experienced No Change in Interest
■ Survey Question:
“I would take additional CS courses if given the opportunity”
IUPUI
UA
Student Interest in CS
■ Disappointing results: neither pilot showed a significant pre-post
impact
■ Neutral results: No loss, but also no gain
■ IUPUI: One change made since reporting these results was to provide
CSP students with course pathways subsequent to CSP completion;
we are monitoring impact, if any
LESSONS LEARNED
Kudos to the CSP Community
■ Fantastic community support for CSP resources
– CS10k Community, Endorsed Providers, Many NSF projects (e.g.,
TeachGlobalImpact.org)
■ CSP curriculum is indeed engaging and welcoming
■ Recommended cooperative learning strategies worked well at both
universities
University-level CSP Course Challenges
■ Curriculum coverage in 1 semester is impossible
(48 contact hours vs over 180 in AP CSP)
■ Culture challenge to faculty committed to a lecture
approach, but opportunity to map back success
from CSP to traditional CS curriculum
■ Covering the programming Big Idea in a single
semester (~12 hours) with non-majors
CSP
CS1
Student Voices
■ Group projects
■ Socially Relevant
■ Wrote programs that they could show off on a cell phone
■ Myth-busting: “I did it” “This is fun!” “Who knew Computing was
Creative?”
■ Pseudo- code
■ Debugging
■ Journaling (however, seems more popular for AP CSP)
Conclusions
■ Summary evaluation results of 3-year student of CSP course at two separate
university College Board Pilots
■ Two separate pathway approaches for CSP
– UA: SEMA students, IUPUI: general math elective
■ Positive Outcomes
– Very deep increase in enrollment diversity with respect to gender
– Significant gains in student self-confidence in CS
– Substantial decrease in DFW rates compared to CS1
– UA: Math Education students increased engagement with CS department and
pre-service training of future CSP educators
■ Surprising Results
– Decrease in URM enrollment at UA, increase at IUPUI
– No change in student CS interest before/after CSP
– Challenges in implementing full AP CSP curriculum in university course
Acknowledgements
■ We would like to thank the College Board, through support from the
National Science Foundation, for selecting us as CS Principles Pilot
instructors.
– In particular, we are grateful to Lien Diaz at the College Board,
Owen Astrachan and Amy Briggs (NSF PIs), and Jan Cuny (NSF).
■ Special thanks to Kathy Haynie and Sheryl Packman for their help with
evaluation data.
■ Part of the project at the University of Alabama was supported in part
by NSF CE21 (1240944).
■ The CS104 course at Alabama also benefitted from the support of
Amber Wagner, Brian Eddy and Mokter Hossain.
QUESTIONS /
COMMENTS
Thank y’all!