Cost Analysis of Online Courses

John Milam

University of Virginia

Introduction

As institutions witness the phenomenal growth of virtual universities, cybercolleges, distance learning programs, and web-based courses, how prepared are they for making the decisions about resources which confront them?

How much does it cost to develop an online course and what is the “break-even” point in enrollment? Is a distance learning program cost-effective and how should this be measured? How much effort and funding should be put into course development? Even if a program loses money, how much is it worth for the institution to be seen as a leader in the use of technology? Does it take less faculty time to run an online course? What are the computing support needs for these courses and does it make sense to pool resources across departments or schools? What is the best use of scarce resources to move into this arena? Is it cheaper and more efficient to use a vendor such as University On-Line or e-college.com to develop the courses?

These are some of the many questions which may be addressed if institutions build a resource allocation model and planning process for developing online courses. An entirely different, but related, set of questions needs to be addressed in terms of assessment. While there are many worthy topics for this type of inquiry, the specific focus of this paper is on cost analysis of online courses.

Of course, the topic of online courses is not the only pressing question which suggests the need for a cohesive planning process and better resource allocation models. It is, though, perhaps the most costly to ignore; not only in terms of cost-effectiveness in using resources, but more importantly in taking advantage of the small window of opportunity which is now available for institutions to develop an online presence. There is fierce and aggressive competition from all sectors in the online market for higher education. Without the types of cost analysis described in this paper, colleges and universities are ill-prepared to meet this challenge and may find themselves left out of the race to capture the huge but finite market of online enrollment.

Literature Review

Unfortunately, while there is a growing literature on how to build web-based courses, develop online degree programs, and incorporate best practices in instructional design, there is much less written about assessment and the success of these courses in meeting student and academic needs. Even less is written about the resource allocation issues which are involved, such as faculty workload, comparing costs between traditional and online courses, and computing support.

A new resource is now available for planners in the Flashlight Cost Analysis Handbook (Ehrmann and Milam, 1999), published by the Teaching, Learning, and Technology (TLT) Group, an affiliate of the American Association for Higher Education. This is the most recent in a series of online and print tools in the Flashlight Program, started in 1993 as part of the Annenberg/CPB Projects to “help educational programs understand and shape the consequences of their own uses of computing, video, and telecommunications.”

The Flashlight Program includes: (1) the Current Student Inventory (CSI), a 500 question/item bank and tool kit for survey development about the use of technology in teaching; (2) the new and evolving Faculty Inventory, of which Part I helps users compare faculty and student perceptions about the use of instructional technology; (3) Flashlight Online, a web-based tool for managing the CSI; (4) the Flashlight “Tool Series,” involving subscription-based access to new tools and updates; (5) the Flashlight Network, for working with institutions and organizations in a wide range of activities such as Roundtables; (6) a monthly email newsletter, F-LIGHT; and (7) the Cost Analysis Handbook with its economic model.

In addition to a detailed discussion with examples of how to build cost analysis models, the Handbook provides advise about many of the organizational constraints which emerge when attempting this kind of resource allocation process, such as team development in model building, shared perceptions of project outcomes, and focusing on the unique cost drivers in a particular institutional setting. Aimed at all types of uses of technology in teaching, the Handbook is not focused primarily to online courses, but is easily adapted for this purpose.

This type of inquiry is also well served by an understanding of the more general costing literature in higher education. The NCHEMS cost of instruction model has been in place for almost thirty years, and is perhaps most useful today, when institutional research offices have the tools to build online data marts and data warehouses which merge financial and student datasets. The heart of the NCHEMS model involves the use of course enrollment data in an induced course load matrix (ICLM), first championed by Suslow (1976). The ICLM, with its focus on departmental consumption and contribution ratios of student credit hours (SCH), is equally valuable in modeling the enrollment of traditional and online courses. The NCHEMS model is still used by some state higher education executive offices (SHEEOs) to calculate discipline-specific staffing ratios, based on student credit hour productivity.

Similarly, the rich literature of the National Association for College and University Business Officers (NACUBO) for indirect cost recovery and micro- and macro-costing is invaluable to this effort. This includes Cost Accounting in Higher Education: Simplified Macro- and Micro-Costing Techniques (Jenny, 1996); A Cost Accounting Handbook for Colleges and Universities (Hyatt, 1983); and the NACUBO Handbook: College and University Budgeting (Meisinger and Dubeck, 1984). Another useful monograph, prepared by a consortium of institutions and organizations in Great Britain, is Management Information for Decision-Making: Costing Guidelines for Higher Education Institutions (Joint Funding Councils, 1997).

While models may be built without attention to these standards in cost analysis, they will fail to address some of the complexities of enrollment, space utilization, departmental and institutional overhead, and faculty workload. This is not to say that only complex models should be built, but that any model development needs to be done with a clear set of assumptions. If assumptions about revenue, for example, will be ignored because student tuition estimates and financial aid data are not available or are too complex to be analyzed within the timeframe and staffing level, this needs to be recognized as a limitation of the model.

Most important to take from the cost literature are the concepts of: (1) the ICLM for enrollment and departmental consumption/contribution; (2) space utilization and allocation costs; (3) revenue stream based on tuition and fees minus financial aid and tuition discounting; (4) faculty workload; and (5) administrative overhead at the department, college/school, and institution-wide levels. The reader is referred to Jenny (1996) for a more complex treatment of these concepts. Jenny provides specific examples for costing at the course level.

Methodology

The Flashlight cost analysis approach is being implemented in different ways at different institutions. The Handbook includes four case studies, each of which involves a unique resource allocation question and an individualized version of the model. However, the purpose and basic steps are the same.

The focus of the methodology is on activity based-costing (ABC). ABC is a way of looking at costs that cross the traditional building blocks of institutions, for which a specific account or set of accounts are inadequate to track expenditures and revenues. As used at institutions such as Indiana University and its campuses such as Indiana University-Purdue University Indianapolis, ABC is part of a larger approach called Responsibility-Centered Management (RCM). Units at all levels, from departments to colleges, are held accountable for the cost of activities for which they are responsible, even if these activities cross the boundaries of departments, accounts, and majors.

ABC is different from traditional accounting, which looks at expenditures and revenues within one or more account structures, based on organizational reporting relationships inherent in the financial chart of accounts. The type of cost analysis is performed by sponsored research and fiscal affairs offices is indirect cost recovery. Administrative costs are shared across research activities based on an agreed-upon indirect cost rate, which is itself based on complex weighting and allocation schemes.

Since sponsored research and service activities use institutional facilities, consume utilities, and require administrative oversight with functions such as human resources administration, it is important to consider these costs for sharing. In order to calculate “hidden” versus direct costs, micro-costing models have been developed. Where the Flashlight model incorporates only minimal calculations of hidden costs, some implementations such as that done by the author at George Mason University are more complex in their use of administrative cost-sharing. This discussion of the costs of online courses will involve some relatively simple calculations of indirect costs (but still difficult to gather data for). The reader is again referred to Jenny (1996) and the NACUBO literature for a more full discussion of micro-costing.

The Flashlight economic model involves seven basic steps. As stated in the Handbook, these are:

1 Identify your resource concerns and the specific questions you want answered.

2. Identify your outputs.

3. Identify the activities that are required to produce your outputs.

4. Identify the academic and support units that participate in these activities.

5. Identify the resources these units consume in their activities.

6. Calculate costs for these activities.

7. Tally the costs of all activities to arrive at your output costs
(Ehrmann and Milam, 1999, p. 11).

For the purposes of this discussion of online courses, three sub-steps are added. Step 5, “identify the resources,” is broken into two – for direct and indirect costs. Step 6 is also broken into two, with additional data about enrollment via the induced course load matrix. A new step is also added to calculate revenue stream based on enrollment, tuition and fees, and financial aid data. The revised steps of a cost model for analyzing online courses now include the following:

1. define the resource issues
2. choose outputs and performance measures
3. document activities and tasks

4. gather faculty and staff workload data

5. collect data on direct costs

6. calculate data on hidden, indirect, or shared administrative costs
7. gather data on enrollment with the ICLM

8. calculate results for each activity

9. calculate revenue stream

10. summarize the results

Steps for a Cost Analysis Model of Online Courses

1. Define the resource issues

Central to any model development is the need to define the objectives. While we are focused on examining cost issues for online courses, there are dozens of complex and informative models. It is impossible to create a single model which will meet a wide variety of needs. Planners are well served by establishing the resource allocation issue first, then building the model. Once a model has been developed and used for a specific question, new questions will arise. One needs to go through the model building exercise again, rather than look for ways to simply tweak the existing model. While this may seem a luxury of time, it prevents a type of planning myopia which sees only the solution, sometimes without truly documenting and challenging the assumptions behind it.

The Flashlight Cost Analysis Handbook recommends stating the resource allocation question directly in a few sentences. For example, that an institution wants to know whether it is cheaper to offer an online section of a course than its traditional counterpart. Cheaper will be defined as sustaining higher enrollment without additional cost.

In examining online courses, many questions beg comparison to their traditional counterparts and this needs to be built into model development from the beginning. An online course may not have an on-campus equivalent. It may be necessary to benchmark against certain activities which make up the course, such as course preparation time or working with students outside of class.

In choosing a resource issue to study, pick something which lends itself to being broken down by activities and within these tasks. If an institution outsources online course development to a vendor, it may be difficult to benchmark the costs by activity to a traditional course. A single fee per enrolled student may be paid per semester to the vendor. While the vendor supports the web server setup, course development, and administrative tasks, these activities are often not broken out into data on time or effort for an institution. Only the activities which are under institutional control can be evaluated.

Whatever the question, planners need to ensure that building a model will benefit the institution by answering resource allocation questions of interest. The Flashlight Handbook suggests that a mock version of a model be created with simulated results. Does the report tell administrators what they need to know to make decisions? If not, it is better to reframe the scope of the model at the outset, rather than have to redefine it after much time and ownership have already gone into its development.

2. Choose outputs and performance measures

In the discussion of resource questions, an example question was framed: whether it is cheaper to offer an online section of a course than its traditional counterpart. The next step is deciding how to measure costs. What does it mean to be cheaper? Is this the overall cost of offering an online course per enrolled student? Per student credit hour? Are costs calculated per semester or over time? This is an important point, since it often takes several semesters to develop an online course. Should the results somehow be amortized to reflect true costs and benefits?

Lessons learned from the four Flashlight case studies described in the Handbook suggest that one of the standard performance indicators, student credit hours, is inadequate for measuring costs in use of technology in teaching. This is because many implementations of a technology are used for less than the traditional fifteen week course. It is this very flexibility in course scheduling, with differentiated stop and start dates and the ability to tailor course objectives to the audience, which makes online courses so appealing. The calculation of credit hours in a fifteen week semester is a traditional mindset which may or may not meet the needs of the resource allocation question of interest.

An alternative presented in the Handbook is weekly student course hours, the number of hours in which students are expected to be engaged in directed, course-related learning activities (different from time spent on homework assignments). For the traditional course which meets three hours per week for fifteen weeks, this equates to forty-five hours per student. Comparable to the measure of weekly student contact hour which is used for calculating space utilization, this measure is more helpful in measuring use.

For models of online course costs which incorporate an assessment component as well, planners may wish to use performance indicators such as cost per passing student or cost per retained student. An online course may potentially meet the need of many more students. How many of these students remained enrolled after the tenth class day? Is there a higher drop-out rate than the traditional class section? Is there a lower pass rate? Or a lower grade distribution? It is impossible to expect that this analysis of course costs will be done in a vacuum which does not take into account assessment issues. While this is not the intent of the model, performance indicators should be chosen which address the true questions of interest.

Since it takes significant time to bring a concept for an online course to fruition, measuring course costs over a single semester may be misleading. You may want to “amortize” course costs over several semesters or years. Similarly, while it is interesting to examine expenditures, this singular focus fails to take into account the revenue side of the equation. Complex models may be built which incorporate revenue data based on student enrollment, calculated from tuition and fee charges by residency and student level and incorporating standard models of financial aid and internal tuition discounting. The performance measure of revenue per online course is misleading if part of the revenue is made up of internal funding with tuition discounting.

3. Document activities and tasks

As the central feature of activity-based cost models, the documentation of activities and tasks for offering an online course is critical to a model’s success. This requires a certain assessment approach which examines the nature of the teaching process and breaks the faculty role into parts. A traditional class may involve activities such as preparation, teaching, and administration. While these may be easily translated into specific tasks, such as creating the syllabus or grading exams, it is this definition of work and its equivalent expression in costs which is at the heart of any costing model. Also, the responses are never as straight forward as one might assume.

In determining the activities and tasks involved in offering an online course, faculty and staff involved are the best source of data. It is helpful to ask what seem on the surface to be simple, open-ended, and innocuous questions. What do you do in this course? How do you handle the same types of tasks you do in an on-campus class?

Working to gather data on the many tasks which make up faculty work, one quickly learns that there is little written documentation of effort. Neither faculty nor staff keep logs of their activities. During an interview, they may think for the first time of the tasks they performed to offer a course. Such planning can be very haphazard. Faculty members may introduce technology into their classes over a period of time. Online courses do not suddenly materialize into cyberspace. Faculty may have been introducing listservs, email, and HTML documents into their classes for several years. A specific course may have evolved to the point where so much Internet-based technology is used that it makes sense to move it entirely online, but this is not necessarily a linear or rational process. In understanding the development and impact of online course development in an institution, one should not be fooled into simplistic models of organizational theory, faculty development, or curriculum design.

Here, as part of model development, there is a great by-product of this process – documentation of teaching roles. If planners are able to document the many tasks and range of activities which make up the faculty and staff support role, they will find that this a itself an interesting piece of data. What exactly is the faculty role? How is the faculty role changed with offering online courses? These are not simple questions. They are at the center of costing models. Institutions such as the University of Phoenix rely solely on part-time faculty. There is no attempt to perpetuate the traditional faculty role. Should schools create an instructional development function to share and pool resources for online course development? This is the model of the vendors such as UOL and ecollege.com (formerly Real Education).

4. Gather faculty and staff workload data

While it may seem difficult to define the activities and tasks which go into developing and offering an online course, it is the gathering of actual workload data which is the most frustrating for planners. No faculty logs are kept. Staff members do not sit with a list of their many projects, so that they can readily tell you how much time they spent setting up a web server and what proportion of this effort was devoted to the specific course. These data are not kept because there are no rewards involved in keeping them. Resource allocation models rarely work with more than the most gross approximation of faculty time.

This is not to say that some institutions do not have faculty workload surveys in place or that department chairs do not have some degree of control over their faculty members’ time. Yet these are accountability controls, not intended for gathering actual data but for monitoring patterns of activity over time.

As part of your model, you may wish to ask faculty to complete a workload survey which documents the time they spend on various activities. You may want them to keep a log over several weeks to get actual data or you may be willing to use estimates. In conducting interviews with faculty about their online courses, you will be surprised to learn that they do not have a clear sense of their time. Or it may be influenced by the activities of the week you interview them. This process of reflection may be as beneficial for them as it is for your model.

One recommendation from one of the case studies in the Flashlight Handbook is that institutions rely on a set of assumptions to guide the use of workload data. For a research or doctoral institution, faculty are expected to teach at least two courses a semester, spend 25% of their time in departmentally-sponsored research (unless they have external funding), and spend another 10% or so in administrative duties such as advising and attending faculty meetings. This suggests that only 65% of their time and salary is devoted to instruction-related activities. If faculty teach two courses per semester or four per year, then this 65% must be distributed among each of the four courses. See the National Survey of Postsecondary Faculty (NSOPF), conducted by the National Center for Education Statistics (NCES) for a source of national benchmarks in faculty workload data (Kirshstein et al, 1997). The NSOPF survey documents the many ways to collect complex workload data.

Another alternative is to ask each faculty member to estimate the proportion of their instructional time devoted to each course they teach. Without data from logs, this is simply an estimate, better informed of course with a listing of typical course activities and tasks to remind faculty of just how many different roles they must juggle in the course of an academic year.

Whether it is with actual logs, estimates, or general assumptions, faculty workload data are essential for any model of online course costs. Similarly, there are many staff support roles and graduate teaching assistant roles which need to be included. For example, if a department secretary is involved in collecting grades and distributing grade reports for a traditional course, how is this handled for an online course? Is there decreased cost because grades are posted online and given directly to the registrar? If these types of differences in administration and support are not somehow taken into account, the calculation of online versus traditional course costs will be misleading.

Using the faculty interview as a source, it is important to find every interaction with support staff which influences course activity and to build this into the model. This includes such tasks as computing staff support for setting up space on a web server, library staff time spent teaching a faculty member HTML in order for her to put up online documents, and departmental evaluation and approval of the electronic or print syllabus.

5. Collect data on direct costs

Direct costs such as faculty compensation, including salary and benefits prorated based on workload, are relatively easy to obtain from the institutional research, budget, and human resource office. It is important to verify these data with the faculty involved. Expenditure data are less useful, not because they cannot be broken down by transaction for actual course expenditures, but because there are often so few direct purchases for a specific course. In one of the case studies described in the Flashlight Handbook, faculty reported bearing the cost of many small items themselves. It is important to include these costs.

Departmental account reports are a starting point. When meeting with faculty to discuss any direct costs that may be related to their breakout of activities and tasks, the department account report is only useful to a degree. Most are broken down into object code, clusters of expenditures based on standard types such as travel or printing for budgeting and allocation purposes. A traditional class may involve $100 worth of copying, but this is not included in any meaningful way in an existing report. An individual faculty may have a code used by the copy center and may estimate that copies made on a certain date were probably for a class. More likely these reports are by account code and give no useful insight into costs. This reality of reporting suggests that planners should put data collection instruments in place before gathering data for their models. Give faculty members a worksheet to keep of personal and departmental expenditures for the online class. Whenever possible, probe for places where there may be missing data.

6. Calculate data on hidden, indirect, or shared administrative costs

In its report Straight Talk about College Costs & Prices, the Congressionally-sponsored National Commission on the Cost of Higher Education discusses the “veil of obscurity" which hangs over financial data. This is perhaps no more true than in regard to indirect or shared administrative costs.

Direct expenditures only capture so much data. Yet offering an online course in department X may not cost the same as if it were offered in department Y, solely because of departmental overhead. Student enrollment is the product, however measured. All instructional activities exist in order to support enrollment. All instruction-related costs must somehow be taken into account in building a cost model for the online course.

For example, department X may offer 50 different course sections in a semester, and department Y offers 40. The departmental staff includes a chair and a secretary. Neither of these staff may spend any time directly related to the online course, but they support the activity in an indirect way via monitoring, oversight, personnel matters, etc. Somehow these costs need to be distributed among instructional activities. This may be done with various allocation schemes, such as by student credit hour or by individual course. Since the overhead of monitoring the success of enrollment is often at the course level, online course costing models may want to allocate departmental costs by the course.

Therefore, department X’s chair and staff compensation costs are allocated to each of the 50 courses, department Y’s to its 40 courses. If the chair and staff salaries are relatively equal, department X will end up costing less per course in administrative overhead. Similar allocation schemes need to be built in at the dean’s level and at the institutional level. These data show that not all departments and schools are equally prepared to bear the cost of an online course. It may be much more economical in terms of overhead to offer an online course in English, where there are many courses and a relatively thin departmental support team, than to offer courses in Anthropology, where there are far fewer courses. The circumstances are always different with different disciplinary structures at institutions. The point is that these support costs are real and need to be included in resource allocation models.

In addition to administrative overhead, which is treated much more completely in Jenny (1996), Hyatt (1983), and Meisinger and Dubeck (1984), there are other hidden costs to consider with online courses – including space and computing.

Much is made of cost savings realized because online courses do not require bricks and mortar. While it is true that classroom space is not required, classroom space is not the largest component of facilities room use, simply the most recognizable. Faculty office, instructional support, clerical support, and computing facilities also need to be included in the use of space by the online course.

In determining the cost of using space on campus, there are many models. The National Cost Commission report suggests the need, based on the work of Gordon Winston, to include opportunity costs in examination of tuition price and subsidy. One possible model is to express opportunity costs as the lost revenue which could have been earned if an institution had rented the space used by the course at fair market value. This calculation also builds in some of the cost of depreciation and maintenance, other facility costs not often taken into account. Regardless of whether an online course uses classroom space, models of online costs versus their traditional counterparts need to include hidden facilities costs. Many traditional courses will have the same or greater costs, so it may be a wash. However, no course costing model is complete which does not include some assumptions about space utilization. If anything, it is important to use these data to show that online courses provide substantive savings based on their decreased use of space, a claim that while intuitively true needs data to support it.

Other computing support costs should be included and it is here that the planner may find the greatest dearth of data or meaningful allocation scheme. An online course uses a listserv. During a given semester, there may be 150 listservs in operation. What is the course’s share of this cost?

A series of questions follow. What server is used to house the listserv? What did the server cost to purchase or upgrade and what does it cost to maintain in terms of utilities and staffing? What activities does the server support and how might these be prorated to allocate the server costs? Each technology used in online courses should be scrutinized in this manner. This includes email, listservs, newsgroups, telnet, threaded discussion rooms, chat rooms, MUDs and MOOs, administrative information systems, web servers, Java simulations, collaboration software, audio/video delivery, and database software such as Perl, Cold Fusion, or Active Server Pages, etc. In many cases, planners will find that there are no existing studies which document the costs of any of these services. They are all ripe for activity-based costing. Faced with this problem, rather than build incomplete models or models that rely on estimates that are little supported with data, recognize that your model adds value by beginning to help document that these hidden support costs do exist.

Of course, if faculty set up their own web server, or if the use of technology is a first in setting up services for an online course, the costs are more readily identified. Be wary of making a single course bear these costs, though. Amortization schemes are in order which spread these costs over time. Online courses do not suddenly build themselves. Pay attention to the evolution of technology in a department’s courses and the ways in which this lends itself to online course development. Resource allocation models need to somehow incorporate this development process.

Other computing costs to be included in the model are the computer, printers, and network used by the faculty, staff, and/or graduate teaching assistant in the course. Some estimates of use may be made, perhaps comparable to the estimate of faculty workload. For example, 65% of activities may be instruction-related, therefore 65% of the PC is instruction related. Of this same 65%, the faculty member teachings four courses per academic year. In this case, 16% of the amortized cost for purchasing the PC and 16% of direct software and other computing costs may be allocated to each course.

7. Gather data on enrollment with the ICLM

In understanding costs, most performance indicators are built around enrollment. These include weekly student contact hours, passing student, student credit hour, headcount, full-time equivalent student, and other calculations. Much is learned by knowing class enrollments, but this is only a starting point. If revenue is to be considered, the induced course load matrix or ICLM must be part of the model.

The administrative information system structure of data used by higher education traditionally includes one record per student per course. Therefore, if a student takes five three credit hour classes, whether or not they meet online or on campus, there will be five records created in the student database. Since one of these is for a specific course, it is easy to calculate that 20% of the revenue generated by a single student’s tuition and fees should be counted for the course.

This simple example is repeated many thousands of times to estimate what proportion of all students course activity goes to the single online course being studied. If there are 100 students enrolled in the online course, and they all take 5 courses each, then course serves the equivalent of 20 full-time equivalent (FTE) students. If the average full-time tuition and fees charge for a semester is $2,000, it may be reasonably estimated that the course earned $40,000 in revenue.

The ICLM has many more benefits than this type of calculation. Most useful is the measurement of departmental consumption and contribution. A department offers classes that are taken by many different students, not just its own majors. When non-majors take its courses, the department is contributing to other departments. When performance measures are based solely on majors, the department suffers since it has a service role that is not being taken into account. Similarly, a department’s majors take many courses outside of the department, making it a consumer of other departments offerings. This balance of consumption and contribution in credit hour course activity is the battleground for departmental resource allocation.

In estimating the cost of online courses, do not fail to address the question of whether the online course contributes to the department’s role in consumption or contribution. If a department continues to serve more outside students, this should be part of resource allocation decisions. Such an argument may help a department get funding for an additional faculty position. The flip side of this perceived service and consumer role is faculty workload. With online courses, it is possible to serve more student enrollment with existing resources, changing the departmental dynamic. It is important to balance the competing tensions of service, consumption, faculty workload, and enrollment when analyzing online course costs. The ICLM is a very useful tool for analyzing these tensions. In the past, it has been very difficult to portray the hundreds of thousands of cells and hundreds of possible relationships between units. With web database applications, data marts can now portray these results in a way which can be understood and used by chairs and deans in their battles over resource allocation.

8. Calculate results for each activity

After data are gathered on the types of activities and tasks which are involved in offering an online course, data are collected about faculty and staff workload. These are then prorated against compensation (salaries and benefits). Any direct costs associated with the online course are documented, even those born out of pocket by the faculty member (for future reference). If at all possible, administrative overhead for at least the department level should be estimated. Dean’s level and institutional administrative cost sharing may also be calculated for any instruction-related expenditures. These indirect costs may be allocated in various ways, perhaps most simply by the number of courses within each unit of analysis.

This resulting total cost for the online course is used to calculate the performance indicator of interest. This often begins with total cost per online course and routinely moves to cost per weekly student course hour.

These costs may also be calculated based on each activity, such as course preparation, teaching, and administrative tasks. This is the first of the many cost “drivers” which may be tweaked as planners analyze a model. For example, an online course may involve many more hours in course preparation than a traditional course. In overall costs, the online course may be more expensive for this reason, because it consumes more compensation. The next time the course is offered, however, this activity is not nearly as labor intensive, driving the cost of course preparation down and making the online course more efficient than its traditional counterpart.

Some of the other cost drivers which may be built into the model include: faculty compensation – what if part-time faculty are used to teach the course; using a vendor to handle the online technology; offering the course in a department where there is less administrative overhead; using an equipment trust fund or grant to pay for new computers instead of building these into the course costs; distributing online documents via CD-Rom; and using less classroom and office space. Each of these cost drivers may be built into the model and may be tweaked to get different results. It is necessary to remember this when analyzing the results the first time data for the model are put together. For while it may appear at first that an online course is equally expensive to offer, what drivers might effect the result? If another, less expensive faculty member taught the course – would the cost go down or would it go up because this person had to spend more time getting up to speed on the technology? Does it truly cost less if the class never meets on campus or is it less expensive to meet several times on campus because this builds a relationship between the instructor and the students and improves their retention and grade distribution? These are only a few of the many questions which should be asked of models.

9. Calculate revenue stream

In calculating the amount of revenue generated by an online course’s enrollment, based on the ICLM data, the results also need to be adjusted based on financial aid. Just as the institution’s course file is used to document the total number of classes taken by students in the course, the financial aid database is also necessary to this calculation. However, of all the data files available for institutions, financial aid may be the most complex and most easy to misinterpret.

In the best case scenario, financial aid data for the students enrolled in an online course are used to calculate what actual tuition and fees revenue charge was paid. This is the net of the tuition and fees charges after any internal tuition discounting or waiver which may be in place. For example, an employee may be enrolled in the course. While the person should be included in measures of faculty workload and costs, she or he may have a tuition waiver as a benefit of employment so that there is no associated revenue.

For most purposes, it is adequate to estimate the percent of enrollment by course level with waivers and with tuition discounts. The average tuition discount per student level or course level may be calculated by the financial aid office, perhaps as part of routine reports for admissions guides or the institutional fact book. With this estimate, a modified revenue stream for the online course is calculated that takes into account the very real practice of tuition discounting and is based on the tuition and fees charges per student level and residency of those enrolled in the class (prorated by their total number of courses).

10. Summarize the results

This step brings the cost and revenue components of the model together to calculate the true cost of an online course, using the performance measure of choice. It is here, though, that planners often face a dilemma – how to interpret the results. For every performance indicator may be interpreted in two directions. A lower cost for the online course, the result for example of increased enrollment, may be seen as cost-effective use of technology. If too low, the same result may be viewed as inadequate academic support.

The model results may show that there is a high cost in terms of faculty compensation for the activity of course preparation, but that this will decrease significantly the next time the online course is offered because the faculty member will only have to update web content. Yet where is the displacement in effort going to take place. Will the faculty member then spend more time conducting departmental research? If they are already engaging in the agreed upon percentage of time for this activity, what is to stop them from spending more time in research, especially if there are no data to track their effort? Perhaps the time originally spent on course preparation should be shifted to some aspect of working with students online in order to improve their retention rate or grade distribution. There are many more such questions which arise from model building. Building a model like this raises more questions than it answers, but this is not in and of itself a negative result. Rather, it forces planners and administrators to document and question their assumptions.

Conclusions

It is perhaps in documenting and questioning their assumptions that models may best serve planners. So often, they are met with difficulties in gathering data to support the model. Some of the most important by-products of this planning process are information about the changing nature of faculty roles in online teaching and a better understanding of the cost drivers which impact a particular performance measure.

There is much more to be said about this type of model building. This paper has involved a review of the new Flashlight Cost Analysis Handbook and its approach to models for evaluating resource use in teaching with technology. The Flashlight model has been applied specifically to the topic of online courses and lends itself well to this purpose.

Clearly, one of the greatest hurdles for planners is in gathering the data they need about faculty workload, administrative cost sharing, and hidden and indirect costs. Yet research and practice standards by NACUBO and NCES have much to offer the novice and advanced modeler in this arena. What is most critical to the success of helping institutions prepare for their online presence is a sense of what is possible with resource planning models. By following the seven basic steps of the Flashlight Handbook or the expanded ten steps discussed herein, planners can walk through the maze of complex issues which face them and create models that bring these competing tensions and priorities of academic management to life. We do our institutions and ourselves a disservice if we fail to use this tool at hand to understand the dramatic changes which engulf us as we enter the 21^st century.

References

Brimson, James A. (1991) Activity Accounting: An Activity-Based Costing Approach. New York: John A. Wiley & Sons.

Brinkman, Paul T. (1989). "Instructional Costs per Student Credit Hour: Differences by Level of Instruction." Journal of Education Finance 15 (Summer 1989): 34-52.

Ehrmann, Stephen C. and John H. Milam Jr. (1999). Flashlight Cost Analysis Handbook: Modeling Resource Use in Teaching and Learning with Technology. Washington, D.C.: The TLT Group.

Ehrmann, Stephen C. and Robin Etter Zuniga (1998) The Flashlight Evaluation Handbook , including the Current Student Inventory (1.0). Washington, DC: The TLT Group.

Gaffney, Eileen M. (1988). "Cost and Productivity Analysis for Higher Education: A Look at the Boston College Approach." Planning for Higher Education 16(1):35-55.

Hample, Stephen R. (1980). "Cost Studies in Higher Education." AIR Professional File #7. Tallahassee, FL: Association for Institutional Research.

Harvey, James, Rita J. Kirshstein, and Jane V. Wellman. (1998). Straight Talk about College Costs & Prices: The Report of the National Commission on the Cost of Higher Education. Washington, D.C.: American Institutes of Research, Pelavin Research Center. <http://www.acenet.edu/washington/college_costs/1998/07july/straight_talk.html>.

Hyatt, James A. (1983). A Cost Accounting Handbook for Colleges and Universities. Washington, D.C.: National Association of College and University Business Officers.

Jenny, Hans. (1996). Cost Accounting in Higher Education: Simplified Macro- and Micro-Costing Techniques. Washington, D.C.: National Association of College and University Business Officers.

Jewett Frank, George Connick, and Richard Markwood. (1998) "Analyzing Revenue and Expenditure Streams." In Technology and Its Ramifications for Data System: Report of the Policy Panel on Technology. Washington, D.C.: National Postsecondary Education Cooperative.

Joint Funding Councils. (1997). Management Information for Decision-Making: Costing Guidelines for Higher Education Institutions. Scottish Higher Education Funding Council, Higher Education Funding Council for England, Higher Education Funding Council for Wales. <http://www.shefc.ac.uk/shefc/publicat/others/costing/contents.htm>.

Kirshstein, Rita, Nancy Matheson, and Zhongren Jing. (1997). Instructional Faculty and Staff in Higher Education Institutions: Fall 1987 and Fall 1992. (NCES 97-470). Washington, D.C.: National Center for Education Statistics.

Meisinger, Richard J., Jr. and Leroy W. Dubeck (1984). NACUBO Handbook: College and University Budgeting. Washington, D.C.: National Association of College and University Business Officers.

Sheehan, Bernard S. and Warren W. Gulko. (1976). "The Fundamental Cost Model." New Directions for Institutional Research 3 (1): 53-74.

Suslow, Sidney. (1976). "Induced Course Load Matrix: Conception and Use." New Directions for Institutional Research 3 (1): 35-51.