Everyone involved in higher
education has questions. Students want to know how they’re doing and which
classes they should take. Faculty members want to understand their students’
backgrounds and to learn whether their teaching techniques are effective. Staff
members want to be sure the advice they provide is appropriate and find out whether
college requirements accomplish their goals. Administrators want to explore how
all of their students and faculty are doing and to anticipate emerging changes.
The public wants to know what happens in college and why.
Everyone has questions. We have
the chance to help them find answers.
Learning analytics is about using
data to improve teaching and learning. You might wonder why there’s suddenly so
much conversation about this previously invisible topic. After
all, institutions of higher education have maintained careful records of
student progress and outcomes for more than a century. They have always been ready to provide a
transcript for every student, reporting all courses taken, grades received,
honors awarded, and degrees conferred. Institutional research offices provide
summaries of these student records to campus leaders, accreditation agencies,
and the public. Why learning analytics now?
Two major trends drive the
current emergence of learning analytics. First, data informing teaching and
learning are increasingly extensive and accessible. Second, innovative new analytic
approaches to digesting, visualizing, and acting on these data emerge every day.
What’s special about this
Practical Learning Analytics has a specific goal: to help us
collectively ponder learning analytics in a concrete way. To keep it practical,
we will focus on using traditional student record data, the kinds of data every
campus already has. To make it interesting, we will address questions raised by
an array of different stakeholders, including campus leaders, faculty, staff,
and especially students. To provide analytic teeth, each analysis we discuss
will be supported by both realistic data and sample code.
Who should take this course?
Practical Learning Analytics should provide something for anyone
interested in higher education: current, former, or future students, policy
makers, academic advisors, data scientists, university administrators, ed-tech
entrepreneurs, faculty members, even the curious public.
This course has been designed to
work for a wide variety of audiences. Its structure is modeled on something
everyone can enjoy: a smörgåsbord –
we’re going to treat the class like one big meal. After we set the table, each
guest may wander the room, taking either a small plate or a full entrée from a
series of courses we offer up. When everyone is full, we’ll gather again over
coffee to hear what people thought. There is no defined way to pursue such a
meal – each diner chooses what’s right for them. And there is no defined way to
take this course – every student must choose what’s right for them.
The course will open with a two
week introduction, exploring the landscape of learning analytics in higher
education and setting the table for the main event. This is followed by a four
week meal during which participants may choose among an array of five different
topics, each presented at two levels: a small plate providing a quick
introduction, or a more filling entrée. Those choosing small plates will still have
the opportunity to work with realistic data, analyzing it with code we provide.
Those choosing entrées will make creative contributions of their own: writing
new code for analysis or visualization of the data we provide, perhaps bringing
in data of their own. After this month of exploration, the final two weeks will
feature a concluding coffee. In them, we’ll review what students learned while
wandering through all five courses, share the best things class members
invented, and provide some concluding remarks.
Try searching for “learning
analytics” in the Google Ngram server…nothing comes up!
Our smörgåsbord will include five
major courses, each offered in both small plate and full entrée sizes. Each
course will provide both a realistic data set and a set of example R code which
can be used to conduct the basic analyses we will discuss. Small plate users
will watch a few video lectures about their topic, complete a short quiz on the
content, download the data and R code, and run an analysis to answer some simple
questions. Users who choose the entrée will go further, extending the code in
both instructor-specified and student defined ways. The really ambitious will
repeat and extend these analyses using their own, local data. An introductory
video for each course will outline what it includes and provide some sense of
what users at each level will experience.
To keep the focus on the
practical, the five courses are designed to explore analyses of interest to
different audiences: students, instructors, department leaders, campus-wide
leaders, and course designers.
LA for students: How to become the student you
want to be? Exploring courses, majors, comparing your performance to others
realistically and richly.
LA for instructors: Performance prediction in a
course: up to and including grade penalties, placement analyses, performance
disparities and their correlates, course-to-course correlation
LA for department leaders: Persistence in a
major, first through short course sequences and then from intention to degree
LA for college/university leaders:
Characterizing the student experience, program evaluation – observing
differences and probing impact, capturing more and better information,
comparing the experience of different groups.
LA for course designers: What affects
performance – behavior measurement, establishing the evidence basis for advice,
then acting to affect performance with technological and human behavior change
techniques, putting real-time data to work – early warning systems and