Retention and completion in online learning programs and courses, especially in institutions which practice open admission, has long been a concern. The costs of non-completion – dropout, repeated renewals of learning contracts, course repeats – are expensive for both learners and institutions. A recent study of non-completion in college and university undergraduate level studies in the United States (US) suggests the costs of each non-completing learner can be as high as US$40,000 – taking into account recruitment, student service and support, tuition and related costs. Less than 50% of those who registered in a program of study in the US – whether two- or four-year, face-to-face or online – completed within six years of beginning their studies.
The situation in Canada is somewhat different. Graduation rates are higher. In Ontario, for example, graduation rates (% of a cohort completing an undergraduate degree within seven years, range from 98-100% (dentistry, ophthalmology, veterinary science) to mid-60% (computing science, theology), with most subjects graduating around two-thirds of the intake within this time period. Ontario college graduations rates are lower than the university rates but higher than the college graduation rate in the US, ranging from 62% to 75%.
In online learning, completion rates vary between 70-95%, but fall dramatically when admission to a program or course is open to all who wish to take it. The Open University (OUUK) has a graduation rate of 13% within eleven years of students starting their undergraduate degree. The percentage of students getting to the exam for an OUUK course declined from 80% in 1975 to 50% by 2005 and 30% by 2015. Completion rates are also an issue for many other open and distance education institutions, especially those switching to online learning from other forms of distance education.
MOOC completion rates are also low. HarvardX and MITx recently reported only 5.2% of people who enrol in one of their open online courses earn a certificate – some 4.05 million persons. Yet, if we really look at who these 5.2% are we find the innovative social entrepreneurs from Pakistan, the dedicated teachers in Cambodia, and the tenacious single mothers in the US completing their degrees. Though some MOOCs have completion rates as high as 60% (here and here), such high rates are unusual and are most often linked to micro-credentials and nano-degree programs offered through a MOOC platform.
What are universities and colleges doing to improve both student retention (students’ progress through a course of studies) and completion (students complete the program or course in which they are enrolled) in full- or part-time online learning? Here we look at ten initiatives producing varying degrees of success.
Ten Investments in Retention and Completion
- Using Big Data and Predictive Analytics to Improve Focused Recruitment. Which applicants are best suited to a specific program – which applicants are most likely to complete? Algorithms take the profile of successful students in a specific program, develop a detailed profile of their background and characteristics and then use these profiles as a basis for selection of incoming students coupled with strategies for support for these students, especially in the first year of studies. The goal is to lower the recruitment cost per student while at the same time increasing both retention and completion. The University of Oklahoma made extensive use of these methods to both improve retention and to expand its intake while lowering costs and focusing scarce resources.
- Using Artificial Intelligence with Predictive Analytics to Uncover Patterns Within a Student Population and Predict Outcomes. South Connecticut University is now able to discern the enrolment patterns of students who are the first in their families (including siblings) to attend college and compare those patterns with those who do not have a parent or guardian who graduated with a four-year degree. By tracing patterns of responses to assessment and the timing of assignment submission, the University provides additional support and advice to students it predicts will struggle to complete. The use of artificial intelligence with predictive analytics is in use at Purdue University, University of Maryland, New York Institute of Technology, Marist College and California State University, Chico (read a summary here). Rio Salado College in Tempe, Arizona uses artificial intelligence with predictive analytics, linked to an exception-based student management system (SMS), to alert its advisors of needed interventions aimed at retaining a student, whom it predicts, may soon withdraw or drop out. This enabled the College to secure a three-year graduation rate for credential-seeking students of 42%, compared to 16% for peer institutions. Its first-term and two-year credit success rates (percentage of credit hours completed by a cohort of students) hover just above 70%, which are also in line with the broad peer group. Reviews of these developments are available here and here.
- Changing Course Design – Shifting from a Curriculum-Focused to Student Experience-Centered Design.One barrier to completion is the poor design of some online courses and the poor use of instructional technologies within online courses. Many courses are still created by faculty with little or no experience in creating online learning courses and without the support of instructional design expertise. Evidence shows clearly that student engagement is a powerful and effective predictor of learning outcomes (see here and here). Engaged design adopted by institutions includes the use of games and simulations, using synchronous spaces, collaborative learning, peer-to-peer networks and authentic learning challenges. The University of New England’s (UNE) Business School includes opportunites for authentic assessment in partnership with companies in the student’s own location; students use their own resources, peer networks and the expertise of instructors to solve real-world problems. The Sheffield Hallam University and the University of Cape Town both use social media for peer networking with the aim of increasing engagement. There are also many Canadian examples of this, which can be seen in the Pockets of Innovation series at teachonline.ca.
- Strengthening Instructor Support for Online Learners. Different online learning programs approach the task of instructional support for online learners in very different ways. Some have full-time faculty members teaching a cohort of 25-30 students, while others have part-time adjunct faculty; some have fixed contact hours while others have anytime contact; some have tutors who support a range of courses and the students within them, while others have tutors for each course. What is known is student-instructor interaction, effective feedback on student responses to questions and assignments, encouragement and engagement are all critical aspects of online instructor support and can have a positive impact on learning outcomes, retention and completion. A variety of new models are emerging, which intensify and strengthen these supports-tutor teams at the University of the Western Cape (South Africa), combining synchronous and asynchronous tutoring at a number of California community colleges, the use of artificial intelligence “QnA” systems such as IBM Watson to support tutorial functions at Georgia Tech, and the use of the tutor-cloud as a way of combination of personalized instruction with cloud-based technologies. Student engagement has emerged as the single best predictor of retention and completion (see here and here). Key to retention and completion is the whole idea behind differentiated instruction, where the instructional strategy moves from “one size fits all” to a strategy based on individual learner needs. This is especially evident in the college sector, where a focus on retention is especially strong. For example, George Brown College in Ontario has a focused student success program, though only moderate success resulted from this intervention. Colleges Ontario summarized many of the activities taking place aimed at improving retention and completion rates. An independent review of such efforts conducted by Deloitte was published in late 2017.
- Strengthening Non-Academic Student Supports and Focusing on Early Intervention. A major reason for dropping out relates to fee and cost issues, personal and family relationships and balancing work and learning. For example, Georgia State University (GSU) found many students lost their scholarships and financial supports because of low grades, but did not seek support and assistance to both improve their grades and regain their financial support. By changing the support system and offering a bursary of just $1,000 for each student who maintained a grade-point-average of 2.75 (most scholarships require at least a GPA of 3), the University increased retention and could now provide study skills, memory skills and other skills training needed to regain their financial security. The Open University of Hong Kong developed an i-Counselling and i-Advising system. The system has two modules: (a) Academic Counseling for handling general queries from prospective students about career development, program/course information, and learning modes, and (b) Academic Advisement for dealing with questions from current students on program specifics, study plans, and graduation checks. By automating advising, students can connect and secure support from anywhere and at any time. Career advice, especially related to course choice and links to internships and co-operative experiences, are also seen as important to supporting completion amongst Australian institutions. Support is focused on specific groups of students – e.g. first generation learners. Colleges in Ontario, for example, worked in systematic ways to reduce attrition amongst this group, as this example from Seneca College shows.
- Making Extensive Use of Formative Assessment and Feedback. One form of student engagement is self-evaluation supported by coaching, mentoring and focused feedback – either from an instructor or from a “smart” online system. The most widely used learning management systems have adaptive, formative assessment “engines”, which permit the learner to engage in formative assessment and then receive feedback and appropriate remedial learning support “designed into” the system (for more about adaptive assessment, see here and here). Frequent feedback, according to Gartner, is a critical component of personalizing learning, which in turn leads to higher retention and completion. At Arizona State University (ASU), the University launched a math readiness program in the fall of 2011, using adaptive learning technology. Students work through the program at their own pace, aided by an instructor. The adaptive system uses student data to continually assess what a student knows, remediate any proficiency gaps identified, and reassess student mastery of course concepts, giving each student a personalized learning path. Instructors gain an in-depth view of which students are on- and off-track and why, so they can intervene in a timely way. Instructors also see which concepts students are struggling with across the board, so they can focus class time on mastering those concepts. From ASU’s own data, it is clear students who master entry-level mathematics are more likely to complete than those who do not. Canada is often seen to be at the leading edge in formative assessment and feedback, supported by the creative development of new assessment tools (e.g. within Desire2Learn’s Brightspace LEAP or the open source machine learning supported TAO).
- Creating and leveraging peer support networks. Research reports suggest peer influences, especially in STEM subjects, significantly influence retention. Online learning institutions are exploring the potential impact of peer-to-peer learning networks, peer assessment and peer social media supports with a view to both enhancing learning outcomes but also increasing completion rates. The development of learning communities as communities of practice / interest, for example, the use of peer mentors and coaches (here and here) as well as peer-to-peer assessment (here and here) are all tools aimed at both supporting the learner in their learning while at the same time building a sense of community.
- Micro-Credentials and Modular, Stackable Learning. Many online programs, especially undergraduate degrees, take many years to complete. Many of today’s students are looking for “work-ready” skills, which they can master quickly and put to use while building over time to a recognized credential. The growth of micro-credentials, stimulated in New Zealand by the Government and elsewhere by employer demand, is a relatively new component of the online learning eco-system – part of a movement to shorten the time to completion. EdX, the non-profit founded by Harvard University and MIT to offer MOOCs, now lists 40 “Micro Masters” programs from 24 colleges and universities around the world. In addition, Coursera, the venture-backed start-up which also works with colleges to develop online courses, says it has added 50 new “specializations” (series of courses that add up to a non-credit certificate) in the past year. Based on a competency framework, micro-credentials have value in their own right, but can be “stacked” or gathered into a program of study, which is recognized as a diploma, certificate or degree (for example, at Deakin University in Australia). As competency-based learning experiences a renewal, we can expect a growth in badges, non-credit and for-credit recognition of these qualifications. The expectation is students can better manage the pace of their learning while finding their own route to personal and career success and this aids retention.
- Work-Based Learning, Transfer Credit and Expanding Prior Learning Assessment. Another reason students drop out is their work or family demands leads them to move away from their educational institution, perhaps to another jurisdiction within the same country or to another country. Learner mobility is a major policy concern of many jurisdictions, especially in Europe and the Commonwealth. To facilitate mobility and support flexible routes to completion, colleges and universities developed effective means for transfer credit, prior learning assessment (here and here) and work-based learning accreditation (here and here). These developments are supported by a growing number of recognition agreements, known as transnational qualification frameworks or agreements (for example here and here). While institutions engage in this work, the student who began their studies at one institution may in fact graduate from another – they do complete, aided by these mechanisms. Also helpful to this strategy are changes in so-called “residency requirements” (the number of courses which need to be taken to be awarded a credential at a given institution) – lowering them or removing them can aid learner mobility.
- Assessment on Demand and Assessment Only Credentials. Another reason for dropout and non-completion is the timing of summative assessment, which can have an impact on student performance: the longer the program or course and the period of assessment anticipation, the greater the risk of dropout. This is why some institutions offer assessment on-demand, so students can be assessed when they are ready. Athabasca University permits students to call for a summative assessment at any time during their contract for a course, as does the Kentucky Technical Community College System with the belief that doing so supports self-management of learning. There are more recent developments permitting students to obtain credentials by assessment only – no course work, no tuition, just access to assessment on-demand – including the University of Wisconsin, the University of Michigan and Purdue. It is the founding rationale of Western Governors University and is a growing trend, especially in the US.
Taken together, these ten routes to improving retention and completion are used in varying ways in a number of institutions.
Retention and completion are major issues for colleges, universities and policy-makers for all enrolled students. Low completion rates affect revenues and costs, workloads and planning, loan default rates, student mental health and well-being and other issues, which are both institution specific and system-wide. For those engaged in learning, completion rates matter – they impact the willingness of institutions to sustain and expand their investments, especially in online learning as a delivery strategy.
Evidence of effectiveness of these measures, with the certain exceptions (e.g. predictive analytics) are scarce – a focused, systematic look at what works and why it works is needed to enable the more widespread adoption of effective practice. Case studies are helpful, but benchmarking and comparative analysis would produce better insights into this issue. Also needed, as the Higher Education Quality Council of Ontario pointed out in 2010, is a cost-benefit analysis of various interventions. In the United Kingdom, benchmarking and audits, especially of college student success strategies, are increasingly used.