COMPUTING EDUCATION

Project-Based
Learning
Kevin Gary, Arizona State University

Repeated, sustained team interactions on

scalable complex problems that require require prior knowledge of calcu-
constant synthesis and the application of core lus. The course was excellent and
helped me understand a num-
computing concepts throughout the curriculum ber of high-level concepts, but at
each point where a derivation of
place students on a trajectory toward becoming the details was required, it was
glossed over or I was sent to an
professional software engineers. external resource.
This happens in many areas of

T
computing. For example, I teach
oday’s aspiring software developer has a pleth- mobile and Web system development, and most of my
ora of easily accessible and entertaining educa- upper-division students have some prior knowledge, typ-
tional options including massive open online ically from one of the above sources. But they don’t r­ eally
courses (MOOCs), step-by-step tutorials, week- understand how things work or fit together, and because
end hackathons, maker events, visual programming tools, they have a false sense of security about their skills, ma-
and easy-to-assemble app builders. This “kit-oriented” neuvering students around their shortcomings and onto
skill-building style is far more palatable than 15 weeks of a path toward deeper learning can be more challenging
insomnia-curing university lectures combined with long than working with a “greenfield” student.
nights in the computing lab learning what it takes to do In The Engineer of 2020,1 the National Academy of Engi-
real programming. neering underlines the importance of a well-rounded skill
Not that these resources and learning techniques are set. James Duderstadt2 argues that purely technical skill
bad, mind you; they can build some level of mastery. But sets aren’t sufficient to translate intellectual output to prac-
invariably something gets skipped. Epiphanies at 2 a.m. tical products and services. Put another way, intense, short-
over Coke and pizza aside, computing educators face a real term learning activities may rapidly build technical skills,
challenge: battling the perception that complex problem-­ but these skills have a short half-life, while more durable
solving skills can be acquired through intense, short-term skills are shortchanged. (Hackathons and maker faires are
learning activities. possible exceptions, as their relentless give-and-take over a
short period can build unique communication skills.)
PROJECT-BASED LEARNING Further, short-term, skills-focused learning activities
Over my summer break, I enrolled in a self-paced MOOC usually do an inadequate job of placing those skills in con-
on data science. I was surprised that the instructor didn’t text: a student might master skill X but is unable to assess

98 CO M PUTE R P U B LISH ED BY TH E I EEE COMP UTER SOCI E T Y 0018-9162/15/$31.00 © 2015 IEEE

Authorized licensed use limited to: b-on: Universidade de Coimbra. Downloaded on June 11,2025 at 09:17:37 UTC from IEEE Xplore. Restrictions apply.
[Link] 98 8/21/15 11:00 AM
 EDITOR ANN E.K. SOBEL
Miami University; sobelae@[Link]

tools and techniques that address the Today’s computing educator also this by ensuring sustained, long-term
same problem space differently. The needs to teach not only how things participation.
professional software developer is of- work “under the hood,” but the process
ten contextually challenged; he or she by which a software engineer arrives at PBL AS THE CORE OF
works for a “Java shop,” a “.NET shop,” a solution. How do I adapt and trans- A COMPUTING DEGREE
or a “Ruby on Rails shop.” And although late my experiences to the problem Most software engineering degree pro-
the numerous reasons for these tech- at hand? How does what I’ve learned grams leverage PBL for various pur-
nology stack decisions go beyond the apply to this situation? This empha- poses. At the program level, a senior
scope of this article, undeniably one sis on process requires the aspiring capstone project focusing on synthesis
of the main barriers to innovation is developer to understand the context is required for accreditation. Many de-
engineers’ inability to see outside the of the problem and its potential solu- gree programs offer a freshman project
boundaries of their own skill set. As tions. Having rote knowledge of a use as a form of orientation and introduc-
computing educators we give students case’s structure isn’t as important as tion to the design process; as an engi-
a mighty hammer, but the nails keep understanding how effective a use neering colleague told me, the project
changing. case−driven approach might be in a gives freshmen time to take all the
Project-based learning (PBL) is an particular scenario. Too often, univer- physics and math they need to know be-
approach particularly well suited to sities outsource this practical aspect fore engineering gets hard. Some indi-
achieve more durable, contextual out- of a student’s education through in- vidual faculty might take the initiative
comes for computing students. Sally ternships or industry-sponsored proj- to use PBL in a topic-specific course, but
Fincher and Daniel Knox 3 describe PBL ects that are supposedly “real-world” few programs require project experi-
as engaging “students in sustained, without defining what that means. A ence throughout the curriculum.
collaborative focus on a specific proj- PBL approach provides “real-world- Sheri Sheppard and her colleagues5
ect, often organized around a ‘driving edness” through a situated learning suggest a layered, integrative degree
question.’” These aspects are critical: process that forces students to identify structure centered on a professional
sustained collaboration over a deep in- key decision-­making factors and draw spine. A PBL spine would enable stu-
teraction requires learners to establish upon prior experiences. Sustained in- dents to contextualize new knowl-
a rhythm and working process to solve teraction over time ensures that stu- edge. Learning to communicate in
a complex problem. dents continuously apply and evaluate context, learning to problem solve in
these experiences. context, and practicing a design pro-
PROCESS, PROCESS, Distinguishing a project process cess in context lead to professionally
PROCESS from a learning process is likewise crit- prepared students.
Software engineering is inherently ical and often gives students trouble. Computing educators need to teach
frenetic. No two projects are exactly I’ve had many complain about earn- to the gaps. For example, software de-
alike, and managing evolving require- ing a grade lower than they expected velopment programs typically offer
ments, expectations, and technologies even though the product their team separate courses in requirements en-
has led to unique innovations such as produced was superior to those of their gineering, verification and validation,
agile practices. Although traditional peers. These students have a hard time or analysis and design to teach the
engineering disciplines are promot- understanding that the ends don’t relevant techniques of each life-cycle
ing “makers” as a new means of un- justify the means. In PBL, fidelity to a phase. However, integrating across
derstanding the design process, soft- project process enables a cooperative these boundaries requires repeated ex-
ware engineers have long recognized learning process.4 Students that hero- periences in a project context. Further,
the need to fluidly transition between ically complete a product the night be- these integrative experiences should
analysis and design, requirements fore it’s due haven’t had a meaningful apply broadly outside of specializations.
and validation, and, more recently, learning experience. Ultimately, the PBL as a pathway to sustained engage-
implementation and deployment (for software they produce is less valuable ment provides a vehicle to synthesize
example, DevOps). Computing educa- than the contextualized learning they computing science foundations—­data
tors must teach across the spectrum acquire through numerous adjust- structures, algorithms, architecture,
of these software engineering process ments and decisions made over the theory—with professional engineering
activities so that students understand project’s duration. Students are natu- principles and specialized topics in the
where they are at all times. rally bursty, and PBL compensates for upper division.

SEPTEMBER 2015 99

Authorized licensed use limited to: b-on: Universidade de Coimbra. Downloaded on June 11,2025 at 09:17:37 UTC from IEEE Xplore. Restrictions apply.
[Link] 99 8/21/15 11:00 AM
 COMPUTING EDUCATION

PBL AT ASU First, PBL benefits from a physical the responsibility not only to equip
For the past decade at Arizona State learning environment that supports our students with the technical skills
University (ASU), we’ve implemented collaboration and creativity, yet space they need to start a career but the abil-
PBL as the central feature of the Bache- availability on most campuses is a con- ity to apply, evolve, and practice those
lor of Science in Software Engineering. stant tug-of-war with administration. skills throughout their lifetime. While
In this curricular design, students en- In addition, PBL can be difficult to as- many forms of learning have their
gage in semester- and year-long proj- sess, both formatively and summatively, place, sustained PBL provides students
ects from sophomore through senior due to the ingenuity some students with these more durable benefits.
years. The PBL integrates software show in hiding behind others’ work.
engineering material with computing Moreover, teaching in a PBL ap- REFERENCES
fundamentals appropriate to the year proach requires the right amount of 1. Nat’l Academy of Eng., The Engineer
of the major: sophomores learn indi- guidance toward learning outcomes. of 2020: Visions of Engineering in the
vidual professional skills through the No two projects or project teams are New Century, The Nat’l Academies
Personal Software Process6 while de- the same, so nobody has the same Press, 2004.
signing and implementing a semester-­ experience—yet everyone needs to 2. J.J. Duderstadt, “Engineering for a
long project emphasizing data struc- achieve the same outcomes. Changing World: A Roadmap to the
tures and algorithms; juniors design Scaling PBL is also difficult both in Future of American Engineering
and implement a year-long project terms of student head count and in- Practice, Research, and Education,”
synthesizing concepts taken from a tegration across the degree program. Holistic Engineering Education: Beyond
focus application area (such as Web, The economics of PBL don’t always Technology, D. Grasso and M.B. Bur-
mobile, or game development) they se- make sense to administration, espe- kins, eds., Springer, 2010, pp. 17−35.
lect at the start of the year; and seniors cially if it needs to be combined with 3. S. Fincher and D. Knox, “The Porous
conduct an industry-sponsored year- recent innovations such as online de- Classroom: Professional Practices in
long capstone project to synthesize gree programs. the Computing Curriculum,” Com-
advanced concepts from their upper-­ Finally, PBL can be exhausting for puter, vol. 46, no. 9, 2013, pp. 44−51.
division coursework. students and instructors alike. PBL 4. K.A. Smith, “Cooperative Learn-
Since implementing PBL, we’ve requires sustained scaffolding and ing: Lessons and Insights from
observed significant changes in our mentorship over a significant period Thirty Years of Championing a
students. Our industry partners note of time: you can’t disengage from the Research-Based Innovative Practice,”
that graduates are better prepared process as easily as you can walk out of Proc. 41st ASEE/IEEE Frontiers in
for professional work—specifically, a classroom when class is over. Education Conf. (FIE 11), 2011,
in terms of adaptability—than peers pp. T1A-1−T1A-7.
from other programs. We attribute 5. S.D. Sheppard et al., Educating Engi-

P
this to the contextualized cooperative roject-based learning isn’t a new neers: Designing for the Future of the
learning experience. Surveys indicate idea, nor does it conform to con- Field, Jossey-Bass, 2008.
that student satisfaction and efficacy sistent definitions of what con- 6. W.S. Humphrey, PSP: A Self-Improve-
are up as well. For example, for the stitutes a project-based experience. ment Process for Software Engineers,
just completed academic year, 22 ju- Although a body of research supports Addison-Wesley Professional, 2005.
niors reported positive attitudes to- PBL in general and its application to 7. ACM/IEEE Computer Society Joint
ward PBL, while only 1 reported a neg- STEM fields in particular, there are Task Force on Computing Curricula,
ative attitude (8 students remained few systematic studies presenting ev- Computer Science Curricula 2013: Cur-
neutral); similarly, 27 of 30 students idence of greater learning outcomes in riculum Guidelines for Undergraduate
indicated PBL was pedagogically computing disciplines. Yet, PBL is typi- Degree Programs in Computer Science,
useful, with no negative responses (3 cally included in recommended sets of 20 Dec. 2013; [Link]
were neutral). The mean for both sur- best practices for computing curricula /education/[Link].
vey responses was also higher than such as the ACM/IEEE’s CS2013.7
any other teaching method utilized, Repeated, sustained team interac-
including lectures, reflection, and tions on scalable complex problems
KEVIN GARY is an associate pro-
problem-based labs. that require constant synthesis and fessor in the School of Computing,
the application of core computing Informatics, and Decision Systems
ADOPTION CHALLENGES concepts throughout the curriculum Engineering at the Ira A. Fulton
Schools of Engineering, Arizona
There are, of course, myriad chal- place students on a trajectory toward
State University. Contact him at
lenges to PBL adoption, both at the becoming professional software en- kgary@[Link].
program and the individual class level. gineers. In higher education, we have

100 COMPUTER  W W [Link] M P U T E R .O R G /CO M P U T E R

Authorized licensed use limited to: b-on: Universidade de Coimbra. Downloaded on June 11,2025 at 09:17:37 UTC from IEEE Xplore. Restrictions apply.
[Link] 100 8/21/15 11:01 AM