MOOCs: Student Motivation, Perspectives and Government Incentives

In his e-Literate article Phil Hill described a classification based on five patterns of student participation in the course (available here).  University of Utrecht Professor Paul Kirschner described three classes of student activity in his business model for online learning (available here).  The Illinois Community College Board [ICCB] had a classification based on the student’s purpose for attending. Subsequent reports used those classifications. (Later the ICCB reports more closely followed the Illinois Board of Higher Education’s RAMP [Resource Allocation and Management Program] Manual and the National Center for Education Statistics instructions which aggregate the statistics treating all enrolled students the same).

These classification schemes can be used to segment the market, organize the learning experience, lead to a business model, and subsequently improve public understanding and federal, state, and local funding policies.

Hill initially used the term “Lurker” as a category. Subsequently “Observers” replaced “Lurkers” and No-Shows added.  The word “lurker” has a negative connotation for those who graduated pre-Internet. I had used the word “browser.”

The U.S. Department of Education, the advocacy non-profits, and several state legislatures are suggesting “completions” as a measure of college or university performance. Some states are suggesting funding should be based on completion rates. Completion and retention rates, using various definitions, are being published from several sources. Most are based on the graduation rates of full-time degree-seeking students.  A college or university could be sharply penalized—even to the point of losing federal financial aid—based on currently published MOOC “completion rates.” Since registered students would be included in the count of students most of the MOOC students would “fail to complete” a degree or certificate reducing the completion rate, now becoming a critical statistic for the college or university.

Which illustrates why understanding the “Washington perspective” is important to avoid miscommunication.

In the 1980s the ICCB developed a Management Information System. The primary objective was to provide data for the legislature. A committee of presidents developed the measures, later approved by the legislative committee. The presidents added a data element to each student’s registration to ask the purpose of their enrolment. Using current terminology, the choices for the registering student were approximately (ICCB declined to make the MIS Manual available):

• Degree or credential seeking
• Career development
• Personal development
• Browsing

Some of examples from those and subsequent discussions may provide some insight into the categories:

• Career development: Integrated circuits, (Lincoln Land Community College District, Illinois for engineers and technicians who need to learn about new developments in electronics)
• Career development: Student A, the bookkeeper or a small local business, enrolls in accounting with no plans for a degree (Waubonsee Community College District, Illinois).
• Career development: Student B, an employee at an Auburn, New York law firm enrolls in a paralegal course. (Very popular and local firms paid for the course).
• Personal development: Justice with Michael Sandel, DVD from WGBH Boston, and possibly a Mozilla badge. (And now a HarvardX course. Every lecturer should watch this for Sandel’s technique).
• Personal development: Auditing “Geography of Wines” at California State University, Northridge (several decades ago; provides wine tasting authorization for students under 21)
• Browsing: Watching Episode 1 of Justice on the web.
• Personal development: Introduction to computers (Amarillo College, Texas, “Bring your computer to our Saturday class and we will show you how to use it”).
• Career development: Cisco IOS Certification (preparation for the external certification examination, Yuba Community College District, California)
• Browsing or personal or career development: Circuits and Electronics MIT Open Courseware, Course 6.002 (or browsing excerpts of the required textbook Foundations of Analog and digital Electronic Circuits on Amazon; the course is available on MITX)

In each of these cases students are not degree seeking, most are taking courses to both advance their career and benefit their employers. All are inconsistent with “completion” as a measure of the quality of a college or university.

These examples illustrate the spectrum of purposes for students to participate in a course. In traditional classrooms there are students enrolled for credit, students who audit, and guests of the instructor. Only students who enrolled for credit are included in federal statistics that “measure” completion (number of degrees from enrolled students). The confusion begins when online courses are offered. These courses can simultaneously serve students with different motivations. The courses also provide a learning opportunity that is not available in the traditional classroom format.

Capturing “purpose” upon registration identifies market segments that have different characteristics and is similar to Hill’s patterns. As the Illinois community colleges did, the MOOC providers need to measure each category separately to make “completion” of degree seeking students more comparable with traditional reporting.

The classifications representing the three perspectives are compared in Table 1.

Table 1 – A comparison of classifications

There is ambiguity for those seeking career development and personal development who are passive. Even though passive participants do not seek a degree there will be benefits to their employers, their community, and the public from their learning.

Approximating Kerschner’s financial model to the ICCB and Hill’s classification, the model can be used to estimate the revenue and costs associated with that model.

Using data from the Duke University report on the bioelectricity course and some tolerable assumptions, the financial distribution using Kirschner’s model is given in Table 2 for a private research university (public colleges and universities would be less). The tuition for active participants is an approximation based on undergraduate tuition and required fees for courses adjusted to a more comparable three-unit course. The tuition for passive participants is based on programs for non-traditional students sometimes labeled community education, extension, or continuing education. Distributing the tuition revenue over the three types of students in the Kirschner model yields an average revenue of $1,041 with 53% of the students paying nothing.

Assuming 5,000 tuition paying students, the total revenue from this model would be would be about $5 million.

Table 2 – An estimate of average revenue per enrolled student

The sum of active and passive participants—47%–becomes a completion rate in the sense participating students accomplish their objective. If degree-seeking students (5%) can be represented by participating in the final examination, the course completion rate is 90%–approximating traditional classroom attendance.

Michael Feldstein commented:

This is a big assumption. Even degree-seeking students drop out before the final examination sometimes in traditional classes.

Furthermore, non-degree-seekers may have reason to take the final exam, so they inflate that number. That said, final exam participation is a good indicator which suggests that completion rates for degree seekers may be a lot closer to traditional classes than we have been led to believe.

Even with these limitations the proposed measure of “completions” should be used for federal and state reporting of online instruction until there is a better measure.

The cost for developing course materials has been estimated at $1 million by Coastline Community College.. The cost of materials is $125 per enrolled student assuming a two-year life and an expected profit of 12.5% on the investment in course development. Similarly HarvardX will prepare course content for participating colleges and universities for $250,000 per course assuming they are provided the equivalent of Open Courseware.

The cost of course delivery is between $.50 and $1.00 per terminal hour. Communications infrastructure to support online students and the computing resources appear to be similar and individually would be half that amount. Students in a 3 unit course are expected to spend 135 hours is class and study; the lower value $68 per enrolled student was used.

Most of the cost will be tutors who assist the online students, monitor group discussions, grade or comment on coursework, and assist with user technology problems. Some commercial MOOCs assume the institution is responsible for course development and all tutorial assistance. That is, the cost for MOOC course delivery can be estimated to be between $68 and $135 per enrolled student with the college or university bearing all other costs. From the perspective of a college or university the word “subsidized” may be more accurate than “free.”

The browsing student who saw at least one video can be estimated at 3 hours or $1.50 to $3.00. A conversion rate of 0.3% browser to enrolled students would recover the cost of providing no cost browsing.

How much tutorial support provided to students varies significantly. Online courses at community colleges have reported one full-time tutor per 100 students for mathematics courses. One full-time tutor per 30 students may be reasonable for some courses. The purpose of using these two values is to illustrate how tutorial costs control the cost of providing an online course. There are, of course, students who take these courses without any tutorial assistance, often self-selected after starting a course.

The cost for tutorial support can be estimated. According to the Bureau of Labor Statistics the median compensation of postsecondary instructors is $1,770 per week. For a 16 week course the cost is $28,320; $40,227 with benefits. Assuming a faculty member would be teaching four courses—a typical teaching university workload, the cost of tutorial support would be one-forth this amount. According to the Chronicle of Higher Education, the average salary for adjunct faculty is $2,987 for a three-unit course and there are no benefits. The resulting costs for 30 students per tutor and for 100 students is shown in Table 3. These costs also include $125 for course materials and $68 for course delivery. “Selection out” may be a result of minimal assistance.

Table 3 Average tutoring cost per student for a ratio of 30 and 100 students per tutor including course materials and computing and communication support.

This shows how important course design and online student services are to costs by increasing the number of enrolled students per tutor. The goal is to reduce student requests for tutorial support. (Another, and hopefully unacceptable, approach is to be very slow to respond to requests so students give up contacting a tutor). For example Rio Salado College has focused their instructional processes and software development to reduce the need for assistance at the same time they have made immediate tutorial support available in order to improve course completion rates. This is one reason for the College’s high retention and low costs).

Knowing both the expected revenue and costs, the net income from an enrollment can be computed. The results are shown in Table 4.

Table 4 Net Income per Enrolled Student

The significant factors that determine net income are pricing—tuition and fees, the ratio of students per tutor, and whether faculty or adjuncts provide tutoring. What is not significant for large enrollments per course are the cost of course materialsy.

The only unit cost that can vary significantly with enrollment is the allocated cost of course materials. If the course had only 1,000 enrolled students the cost of course materials would increase from $125 to $625. Colleges and universities planning to offer an online course should identify the breakeven enrollment early to determine if the course is viable for the anticipated enrollment.

So far the MOOC course offerings have sufficient registrations and enrollments that the Kirchner model suggests could be viable, as it has been for the open universities. Offering some courses “free” may serve to introduce prospective students to subsequent tuition-based course offerings of a college or university and therefore subsidize “free” courses. Data is not available now to justify such a business strategy.

edX’s MIT and Harvard have a different economic situation. The incremental cost of course materials is zero since the courses are being developed to serve the traditional enrolled student. As Harvard President Derek Bok is reported to have said: “There is nothing you can do to prevent these students from learning.” Tutorial assistance may not be needed for HarvardX and MITX students though teaching fellows have been abundantly available for their students.

The data here is only approximate but may illustrate a procedure for evaluating economic viability. It also provides values that can be approximated from available data.

The Kirschner model represents the successful practices of the open universities. It also mirrors the emerging organization of colleges and universities—blended courses that incorporate the benefits of learning technology, offerings through community education for students who want to learn but are not necessarily, at this time, seeking a degree or certificate, and the current practice of making online learning materials—open courseware—available to browsers. This is a more differentiated model of higher education than most pre-Internet college graduates experienced. But as Executive MBA programs have emerged to meet the needs of current workers who did not take an MBA program and now cannot take a traditional MBA program, the same changes to learning are occurring throughout higher education.

This may not be called innovation, but it is transformative and everyone benefits.

Edited by Paul Heald