1. Desire2Learn Student Success System
Guiding Student Success
Using Predictive Analytics to Help Drive the Path to Graduation
Rhonda Gregory
Director of Instructional Technology
Adjunct Instructor
Greenville College
8. As a Christian liberal arts college, we are committed to
transforming lives, so we want to do whatever we can to help our
students be successful in their courses, stay in their program, and
complete their degree. The Insights Student Success System is
part of that commitment.
14. We gained critical understanding of predictive analytics, student behaviors in
the learning environment, and course design, and we gained valuable insights
into our existing procedures for at-risk interventions.
15. We were able to use the live predictions and study the tool configuration in even
deeper detail all while gaining valuable faculty feedback about the accuracy of the
predictions. Again, the process so far has been a very manageable and has
provided a meaningful experience that we feel empowers us to use real and
relevant data.
16. 16
However, one professor was amazed at the information available. He said that if he
would have seen the information in the first few weeks of the class (instead of week 11 or
12 when he first saw the information); he would have realized that one particular student
coming to class was not otherwise engaged in the course. Because that professor doesn’t
use a lot of graded assessment opportunities, he wasn’t seeing the lack of engagement
just in the gradebook. He did see the lack of engagement in the Learning Environment
based on the S3 reports.
17. There are no specific student advisers at Greenville. Instead, faculty wear dual
hats of teaching and advising. However, there is also a Student Success Office at
Greenville who have access to the Student Success System to look for trends
They are our biggest proponents of accessing more data to facilitate success.
18. Valuable insights into procedures for at-risk interventions and learning
environment usage patterns
Faculty immediately saw value in the rich visualizations - particularly the Course
Timeline and Risk Quadrant
19. 19
Baseline Retention and S3 Impact Study
Building out historical data in the LMS
Gathering baseline data from the registrar’s office along with key data points
from the D2L LE
Map gains in retention figures pre- and post-implementation
20. 20
“The unique visualizations in the Insights
Student Success System give us an incredibly
different viewpoint on how
our students our trending – much more than we
can do in our own minds.”
Rick Tanski,
Principal
Academy Online High School
Academy School District 20
Learner Progression and Achievement
Mapping Colorado and Common Core Standards to LE
Competencies , Learning Objectives and Grades Tool
New perspectives on establishing, tracking, measuring
and monitoring learner success
Roll-out to all 25 sections at AOHS
21. 21
“The most important ingredient is that we believed in the Student Success System –
both in its approach to individualize the predictive model for each course and, then,
what it could for our student’s success. We dedicated ourselves to this effort because
of the incredible collaboration with the development team.”
Lorna Wong,
Director, Learning Technology Development
Office of Learning & Information Technology
University of Wisconsin-Madison
23. Initial Deployment
Course simulations and model creation
Develop confidence in predictive modeling
Initial training of staff and faculty
Tweaking, Training and Development
Additional courses and faculty onboard
Developing and documenting best practices
Adoption and engagement with the learning environment
24. Initial Deployment
Course simulations and model creation
Develop confidence in predictive modeling
Initial training of staff and faculty
Tweaking, Training and Development
Additional courses and faculty onboard
Developing and documenting best practices
Adoption and engagement with the learning environment
Next steps for student success
Administrative and procedural reform
Folding into everyday teaching and learning
Initiate student interventions
25. Uses predictive analytics to deliver a real-time, personalized academic progress and mentoring tool
Early guidance that assists students while they are in-flight within their courses
Identify academically at-risk, disengaged or isolated students within the first critical weeks of
semester start
Taps into the data you already have on-site at your campuses
Insights ™ Student Success
System
26. Desire2Learn Student Success System
Guiding Student Success
Using Predictive Analytics to Help Drive the Path to Graduation
Rhonda Gregory
Director of Instructional Technology
Greenville College
Notes de l'éditeur
TY & Good afternoon. My pleasure to be here to share a little bit about the early adoption experience of Desire2Learn’s Student Success System at Greenville College. I’ll also share about the process Greenville has gone through, and is still going through today, to utilize predictive analytics in order to improve student success rates at our institution.
So, what about guiding student success? It’s a very noble proposition and one that is at the heart of every educator I know. It’s what we live and breathe every moment of every day.
As a Christian liberal arts college, we are committed to transforming lives and we want to do whatever we can to help students be successful in their courses, stay in their program, and complete their degree. That’s not very different from what each of you want for your students, and not different from what the other early adopter sites wanted either.
So why is this so important now? What is driving the discourse around student success? Why are we looking for new tools and technologies like predictive analytics to help drive the path to graduation?
Are traditional methods not enough – perhaps even outdated - in the 21st century? Are the needs of our leaners changing so much that it requires a new success paradigm?
To get an understanding of the answers to some of these questions, we need to get at the root of the change driver.
The end goal for all education institutions is to deliver skilled quality graduates into the workforce. College- and career-ready learners if you will.
In fact, the economic future of our nation is rooted in the academic future of our children.
. . . but this is what we are seeing. . .our retention, achievement, and completion rates are at a staggering low and we have one of the most remediated populations in history.
Only 1/2 of all learners who enter higher education will ever graduate.
1/2 of 1st year students are in some form of remediation.
Roughly 1/4 of high students will not move on to higher education.
**Nat’l avg. appx. 72% of freshmen retain to their sophomore year in college.
We have done so much in the last six decades to get students to that front door (i.e. GI Bill, Civil Rights Act, Higher Education Act, etc.) – but we have to stop them from walking back out that door.
So why are we in this predicament? What has brought us to this point?
The situation is compounded by the complex student demographic that exists on campuses across the nation.
Educators today are presented with one of the most diverse student demographics ever known due to the highest numbers of students from lower socio-economic backgrounds, financial aid recipients, first generation learners, adult learners, etc. This situation leads to some of the most complex learning needs in history.
It is a student population in need of complex and sustainable help, guidance, and support – amidst crushing budget constraints, landmark funding reforms, and dwindling faculty resources.
One of the things GC prides itself on is its deep sense of community on campus.
We understand the need to cultivate an environment where our learners can thrive in all areas of their lives: spiritually, mentally, and physically.
It is our goal to utilize the best practical technological strategies available to us in order to help create that type of environment for a student’s education.
Predictive analytics is one of those tools that we, like many institutions of higher learning, are turning to in order to help.
In this 21st century learning landscape, how can we identify and guide the most fragile, at-risk learners?
Recall national freshmen retention rate - approximately 72%?
In ‘12-13 Gardner Inst. Study: GC’s freshman retention rate was just shy of 70%.
This was concerning to us for a number of reasons, and we believe that this failure to retain and later graduate students is a problem that we can help improve if we can identify those students at risk early, and intervene appropriately.
The facts and results of our year-long self-study presented us with a huge challenge and an opportunity to improve:
--Improve the academic future of our students
--Improve our retention, achievement, and completion rates
while continuing to deliver quality education to our students on an ever shrinking budget.
One of the resulting initiatives of our YR-long study w/Gardner led us to the Insights product of Desire2Learn, where we knew existing data in our LMS could help us create the conditions for success.
We felt this crossroads offered an opportunity - one that we would allow us to forge new paths and create new possibilities for where innovation and technology could be implemented to improve student success.
It is an opportunity to build a culture around completion.
There’s never been a better time to use the volume of data that institutions already have on their campuses to provide faculty, students, and student success professionals with the tools & technology they need to drive learner success and deliver college- and career-ready learners into the workforce.
It is time for a fresh approach. Using predictive modeling in combination with dynamic and interactive data visualizations, we at GC, like the other early adopter schools Mid-American Christian and UWS - are beginning to transform our data into powerful insights on key learning trends in their organizations - guiding instructors to quickly identify which learners are academically disengaged or at-risk and that the time is right to take prescriptive action.
To help us create a learning environment where our learners can thrive from freshman year to graduation, we decided to start with a specific plan by focusing on high-enrollment freshman level classes. We are focused on identifying at-risk freshmen earlier in the semester than we had done in the past using traditional methods.
We are using existing learning data deep in the LMS (student grades and engagement information) to help identify which students are potentially at risk for dropping out, withdrawing, or failing altogether.
Using predictive analytics affords us a new perspective on our learners’ progress and success. These insights are also helping us to integrate faculty input into the overall institutional retention effort and other institutional policies.
After some initial tweaking and adjustments during our pilot term, what we saw was remarkable. We gained critical understanding of predictive analytics, student behaviors in the learning environment, and course design, and we gained valuable insights into our existing procedures for at-risk interventions.
During this time, we chose not to initiate any student interventions based on the predictive models because we were still in the phase of learning how to interpret that information accurately.
Even so, early feedback from our faculty indicated that for those Instructors with 50 or more students in a class, a tool that allows them to see students at risk sooner than they normally could identify is particularly exciting!
Prior to using S3 with the initial group of 18 courses in the fall of 2013, we chose 5 similar courses and ran simulations using data from previous terms.
One of the challenges we’ve faced is the absence of a rich historical data set for these courses, as we have just now been using the D2L Learning Environment for our 2nd full academic year.
Frequency of course offerings
A majority (59%) of classes have fewer than 20 students
The predictions get better as more data is available, and we are on the lower edge of that scale right now.
Initial tweaking and adjustments to our environment was common for all adopters. You want S3 to mirror your academic infrastructure.
One of our goals was to have faculty feel comfortable and confident using S3.
We have been able to begin to demonstrate the advantages of the S3 and the LE itself to faculty through consistent and targeted faculty development.
Faculty are able to make sense of the data through straightforward visualizations, such as the risk quadrant plot and course timeline present here.
Click through to show Quote/story
The promise of predictive analytics to deliver early indicators of academic struggle is driving new opportunities for our faculty to understand the engagement, performance, and achievement of our students.
As a small, private institution, GC prioritizes delivering opportunities for students to engage actively with faculty.
The student-faculty ratio at GC is 14:1, no full-time advisors – faculty serve this role.
I should be clear that during this first semester, we were not using the system for the purpose of actually monitoring student performance.
We were focused on training the faculty about using the platform and for tweaking the model based on the data being gathered.
It was important for us all to figure out how best to use the information and for understanding what the predictions meant.
This was also an opportunity for the faculty to take the information that S3 was telling them and compare it with what they were actually seeing in the classroom.
Since there are fewer of the high enrollment freshmen classes scheduled in the spring term, we currently have 12 classes with S3 deployed.
The spring list is our first full-term roll-out of the tool, and it involves additional faculty training. In total, over the first two semesters, twenty faculty members have been involved in the project.
All of our S3 classes so far are F2F & use D2L as a supplement.
Wide variety of usage patterns among faculty
With the deployment and training work going on, we have also begun to develop a set of best practices for using and engaging with the learning environment.
We plan to roll out S3 to some online courses and additional high-enrollment introductory type classes as we continue to grow our course database and LMS usage.
Metrics – gathering info from our SIS & the D2L LE to study as roll-out continues next year
One goal is to determine what changes in retention have happened since deployment, which will be an ongoing cycle of assessment data (HLC connection)
Procedural change: using predictions to check for student progress on a systematic basis – every third Friday of the month.
This check, done by our Student Success office, will provide feedback about student progress to initiate conversations with faculty and struggling students.
The insights we are gaining from the implementation of D2L’s Student Success System are helping us make more informed decisions about intervention strategies with individual students.
Not just for Higher Education – one of the early adopters was a K-12 school in Colorado. Rick and Mike could not make the trek to New Orleans but you can see how S3 is having incredible impact in understanding learner progression and achievement.
Using the grades target to demonstrate learner progression and achievement in a whole light. The grades book is just one optic or perspective. Here we can see how Rick Tanksi and Michael Arsenault are mapping those standards and demonstrating success.
The first early adopter was University of Wisconsin System. They continue using S3 to this day and will speak next.
Sociogram snapshot – of a course discussion in a course in May 2013
See 3 distinct groups of learner communities. See 6 distinct learners (red dots) who are at risk and a few more on the way.
To bring this all together & recap our path…At GC we are still in the process of leveraging this innovative technology to improve our situation and the experience of our students.
GC Early Adopter Phase One-FA13 The roll-out of simulations and model creation was very simple to deploy.
A) Created (5) simulations on already completed courses: gave us a better feeling for the best way to configure our models for current courses.
B) Created predictive models on 18 targeted courses already in progress & trained 14 faculty.
C) Studied the tool - gained valuable faculty feedback about accuracy & usefulness.
D) Choice not to initiate any student interventions
GC Early Adopter Phase Two-SP14
A) Deployed to 12 freshman F2F courses.
B) Continuing our faculty development efforts (tool & intervention plans).
C) Developing best practices for LE & S3 usage for future faculty training.
D) Developing standard procedures for student success office.
GC Phase 3/Next Steps
A) Continuation of best practices & procedural development – and hopefully implementation!
B) Increase faculty buy-in. We are looking for more in-depth adoption on a day-to-day teaching & learning practical level.
C) Continue to work w/D2L on tool reform & excited about possibilities to potentially track students from one course to another.
D) Initiate student interventions as needed and begin to improve student success & retention!
The data accumulating in your learning environment and in the enterprise systems on your campus is a valuable asset.
The day-to-day learner--instructor interactions define the very essence of an engaged learning experience.
S3 – read slide lines 1-4 as they come in
By tapping into that rich historical narrative, you can learn a tremendous amount about the learning experience at your institution and discover simple, yet focused opportunities that improve the learning experience.
Focused on a way to use existing learning data to help identify which students are potentially at risk for dropping out, withdrawing, or failing altogether, predictive analytics technology can dive deep into key achievement, engagement and completion data and to deliver an unprecedented view of learner progress and success.
Using the Insights Student Success System predictive modeling in combination with dynamic and interactive data visualizations, Greenville College, Mid-American Christian and UWS are beginning to transform their data into powerful insights on key learning trends in their organizations - guiding instructors to quickly identify which learners are academically disengaged or at-risk and that the time is right to take prescriptive action.
While it is still early days, the promise of predictive analytics to deliver earlier indicators of academic struggle than traditional methods, is driving new opportunities to understand engagement, performance, and achievement within higher education institutions.