Understanding the Implications of Online Learning for Educational Productivity

Contributed by:
Jonathan James
The purpose of this report is to support educational administrators and policymakers in
becoming informed consumers of information about online learning and its potential impact
on educational productivity. The report provides the foundational knowledge needed to examine
and understand the potential contributions of online learning to educational productivity,
including a conceptual framework for understanding the necessary components of rigorous
productivity analyses, drawing in particular on cost-effectiveness analysis as an accessible
the method in education.
2. Understanding the Implications of Online Learning
for Educational Productivity
U.S. Department of Education
Office of Educational Technology
Prepared by:
Marianne Bakia
Linda Shear
Yukie Toyama
Austin Lasseter
Center for Technology in Learning
SRI International
January 2012
3. This report was prepared for the U.S. Department of Education under Contract number ED-
01-CO-0040 Task 0010 with SRI International. The views expressed herein do not
necessarily represent the positions or policies of the Department of Education. No official
endorsement by the U.S. Department of Education is intended or should be inferred.
U.S. Department of Education
Arne Duncan
Office of Educational Technology
Karen Cator
January 2012
This report is in the public domain. Authorization to reproduce this report in whole or in part
is granted. While permission to reprint this publication is not necessary, the suggested citation
is: U.S. Department of Education, Office of Educational Technology, Understanding the
Implications of Online Learning for Educational Productivity, Washington, D.C., 2012.
This report is available on the Department’s Web site at http://www.ed.gov/technology
On request, this publication is available in alternate formats, such as Braille, large print, or computer
diskette. For more information, please contact the Department’s Alternate Format Center at (202)
260-0852 or (202) 260-0818.
Technical Contact:
Bernadette Adams
Senior Policy Analyst
Office of Educational Technology
4. Exhibits ................................................................................................................................ ii
Acknowledgements ............................................................................................................ iii
Executive Summary ............................................................................................................. v
Introduction.......................................................................................................................... 1
Overview of Online Learning for Secondary Education .................................................... 1
Purpose of this Report ..................................................................................................... 3
Introduction to the Measurement of Educational Productivity ............................................... 5
Estimating Program Costs ............................................................................................... 8
Documenting Context and Implementation .................................................................... 10
Measuring Program Outcomes ...................................................................................... 10
Cost-Effectiveness Research Requirements .................................................................. 12
The Productivity Potential of Online Learning .................................................................... 15
Opportunities to Reduce Educational Costs Through Online Learning ........................... 25
Implications ....................................................................................................................... 33
The Need for Transformation ......................................................................................... 34
Suggestions for Future Research .................................................................................. 35
Appendix A: Additional Resources ................................................................................... A-1
General Productivity in Education ................................................................................ A-2
Productivity and Educational Technology..................................................................... A-3
Key Resources on Types and Prevalence of Online Learning.................................... A-13
Quality Standards for Online Learning Programs ....................................................... A-15
References .................................................................................................................... A-17
5. Exhibit 1: Components of Educational Productivity Analyses ............................................... 7
Exhibit 2: Comparison of Per-Pupil Spending .................................................................... 26
6. This issue brief was developed under the guidance of Karen Cator and Bernadette Adams
Yates of the U.S. Department of Education, Office of Educational Technology.
At SRI International, Marie Bienkowski, Barbara Means and Robert Murphy provided
advice and insightful feedback on earlier drafts of the report. Ashley Lee and Allison Steele
provided research assistance. The report was edited by Michael Smith. Kate Borelli
produced graphics and layout.
The authors made their best attempt to incorporate the thoughtful guidance provided by
reviewers of earlier drafts of this report, including Cathy Cavanaugh (University of Florida),
Fiona Hollands (Teachers College, Columbia University), Kemi Jona (Northwestern
University), Glenn Kleiman (Friday Institute for Educational Innovation), Robin Lake
(University of Washington) and Henry Levin (Teachers College, Columbia University). The
authors are grateful for their constructive comments.
8. Executive Summary
Educational systems are under increasing pressure to reduce costs while maintaining or
improving outcomes for students. To improve educational productivity, 1 many school
districts and states are turning to online learning.
In the United States, online learning alternatives are proliferating rapidly. Recent estimates
suggest that 1.5 million elementary and secondary students participated in some form of
online learning in 2010 (Wicks 2010). The term online learning can be used to refer to a
wide range of programs that use the Internet to provide instructional materials and facilitate
interactions between teachers and students and in some cases among students as well.
Online learning can be fully online, with all instruction taking place through the Internet, or
online elements can be combined with face-to-face interactions in what is known as blended
learning (Horn and Staker 2010).
The purpose of this report is to support educational administrators and policymakers in
becoming informed consumers of information about online learning and its potential impact
on educational productivity. The report provides foundational knowledge needed to examine
and understand the potential contributions of online learning to educational productivity,
including a conceptual framework for understanding the necessary components of rigorous
productivity analyses, drawing in particular on cost-effectiveness analysis as an accessible
method in education. Five requirements for rigorous cost-effectiveness studies are described:
1) Important design components of an intervention are specified;
2) Both costs and outcomes are measured;
As defined in this report, productivity is a ratio between costs and outcomes that can be improved in one of three ways: by
reducing costs while maintaining outcomes, improving outcomes while maintaining costs or transforming processes in a
way that both reduces costs and improves outcomes. Any improvements in productivity are likely to require initial
investments, but successful efforts reduce costs over the long term, even after these initial investments are taken into
9. 3) At least two conditions are compared;
4) Costs and outcomes are related using a single ratio for each model under study;
5) Other factors not related to the conditions being studied are controlled or held
The report also includes a review of ways that online learning might offer productivity
benefits compared with traditional place-based schooling. Unfortunately, a review of the
available research that examined the impact of online learning on educational productivity
for secondary school students was found to be lacking. No analyses were found that
rigorously measured the productivity of an online learning system relative to place-based
instruction in secondary schools. 2 This lack of evidence supports the call of the National
Educational Technology Plan (U.S. Department of Education 2010a) for a national initiative
to develop an ongoing research agenda dedicated to improving productivity in the education
sector. The evidence summarized in this report draws on literature that addressed either costs
or effectiveness. These studies typically were limited because they did not bring the two
together in a productivity ratio and compare results with other alternatives.
Given the limitations of the research regarding the costs and effects of online instruction for
secondary students, the review that follows also draws on examples and research about the
use of online learning for postsecondary instruction. While there are many differences
between higher education and elementary and secondary education (e.g., age and maturity of
students), postsecondary institutions have a broader and longer history with online learning
than elementary and secondary schools. The intention is to use the literature from higher
education to illustrate concepts that may apply to emerging practices in elementary and
secondary education. Findings from the studies of higher education should be applied with
caution to secondary education, as student populations, learning contexts and financial
models are quite different across these levels of schooling.
While rigorously researched models are lacking, the review of the available literature
suggested nine applications of online learning that are seen as possible pathways to
improved productivity:
Two research reports—an audit for the Wisconsin State Legislature (Stuiber et al. 2010) and a study of the Florida Virtual
School (Florida Tax Watch Center for Educational Performance and Accountability 2007)—include data about costs and
effects. These reports suggest that online learning environments may hold significant potential for increasing educational
productivity. Both found that online learning environments produced better outcomes than face-to-face schools and at a
lower per-pupil cost than the state average. However, these conclusions must be viewed cautiously because both reports
lacked statistical controls that could have ruled out other explanations of the findings.
10. 1) Broadening access in ways that dramatically reduce the cost of providing access to
quality educational resources and experiences, particularly for students in remote
locations or other situations where challenges such as low student enrollments make
the traditional school model impractical;
2) Engaging students in active learning with instructional materials and access to a
wealth of resources that can facilitate the adoption of research-based principles and
best practices from the learning sciences, an application that might improve student
outcomes without substantially increasing costs;
3) Individualizing and differentiating instruction based on student performance on
diagnostic assessments and preferred pace of learning, thereby improving the
efficiency with which students move through a learning progression;
4) Personalizing learning by building on student interests, which can result in
increased student motivation, time on task and ultimately better learning outcomes;
5) Making better use of teacher and student time by automating routine tasks and
enabling teacher time to focus on high-value activities;
6) Increasing the rate of student learning by increasing motivation and helping
students grasp concepts and demonstrate competency more efficiently;
7) Reducing school-based facilities costs by leveraging home and community spaces
in addition to traditional school buildings;
8) Reducing salary costs by transferring some educational activities to computers, by
increasing teacher-student ratios or by otherwise redesigning processes that allow for
more effective use of teacher time; and
9) Realizing opportunities for economies of scale through reuse of materials and their
large-scale distribution.
It is important to note that these pathways are not mutually exclusive, and interventions
intended to increase productivity usually involve multiple strategies to impact both the
benefit side (pathways 1–4) and cost side (pathways 5–9).
Determining whether online learning is more or less cost-effective than other alternatives
does not lend itself to a simple yes or no answer. Each of the nine pathways suggests a
11. plausible strategy for improving educational productivity, but there is insufficient evidence
to draw any conclusions about their viability in secondary schools. Educational stakeholders
at every level need information regarding effective instructional strategies and methods for
improving educational productivity. Studies designed to inform educational decisions should
follow rigorous methodologies that account for a full range of costs, describe key
implementation characteristics and use valid estimates of student learning.
Even less is known about the impact of online learning for students with disabilities.
Regarding potential benefits, the promise of individualized and personalized instruction
suggests an ability to tailor instruction to meet the needs of students with disabilities. For
example, rich multimedia can be found on the Internet that would seem to offer ready
inspiration for meeting the unique needs of the blind or the hearing impaired. In fact,
standards for universal design are available both for the Web and for printed documents. In
addition, tutorial models that rely on independent study are well suited to students with
medical or other disabilities that prevent them from attending brick-and-mortar schools.
However, while online learning offerings should be made accessible to students with
disabilities, doing so is not necessarily cheap or easy.
Any requirement to use a technology, including an online learning program, that is
inaccessible to individuals with disabilities is considered discrimination and is prohibited by
the Americans with Disabilities Act of 1990 and Section 504 of the Rehabilitation Act of
1973, unless those individuals are provided accommodations or modifications that permit
them to receive all the educational benefits provided by the technology in an equally
effective and equally integrated manner. The degree to which programs make such
accommodations is not yet known. To address this need, the U.S. Department of Education
recently funded the Center on Online Learning and Students With Disabilities, a five-year
research effort to identify new methods for using technology to improve learning. Similarly,
research regarding the degree to which current online learning environments meet the needs
of English language learners and how technology might provide a cost-effective alternative
to traditional strategies is just emerging.
The realization of productivity improvements in education will most likely require a
transformation of conventional processes to leverage new capabilities supported by
information and communications technologies. Basic assumptions about the need for seat
time and age-based cohorts may need to be reevaluated to sharpen focus on the needs and
interests of all students as individuals. And as a rigorous evidence accumulates around
effective practices that may require institutional change, systemic incentives may be needed
to spur the adoption of efficient, effective paths to learning.
12. Policymakers and educators do not yet have the needed rigorous evidence to answer some
seemingly basic questions about when, how and under what conditions online learning can
be deployed cost-effectively. More research is required to guide the deployment of online
learning to its greatest effect. Research approaches should explicitly consider educational
productivity. Organizational research is also needed to understand the incentives and
barriers to employing the most cost-effective approaches to quality education for all
14. The need to do more with less is an imperative for decision
makers in nearly every economic sector. Education is no Definition of Productivity
exception. State and local education systems face the dual
Productivity is defined as the relationship
challenges of improving outcomes while confronting between program inputs (measured in
budgetary declines. Reducing costs without sacrificing terms of financial value or time) and
quality, or doing better with what is available, requires outcomes and outputs (including both
quantitative measures and measures of
improvements in productivity (see Definition of outcome quality). Productivity can be
Productivity sidebar). increased by
• reducing costs while maintaining
Productivity improvements is one of the primary goals of outcomes relative to other
the online learning systems that are rapidly proliferating in
• improving outcomes while
secondary education. This report is intended maintaining costs or
• to summarize what we know to date about • both reducing costs and improving
productivity as it relates to online learning and outcomes.
• to offer guidance to policymakers who are faced
with the decision of whether and how to implement
this strategy.
Overview of Online Learning for Secondary Education
The available evidence suggests that schools are using information technologies with the
intention of expanding access, improving instructional quality and reducing costs associated
with traditional instruction. Many districts and states have turned to online learning (see
Definition of Online Learning sidebar) to replace or supplement teaching in brick-and-
mortar schools. For example, journal accounts indicate that some schools and districts are
ending traditional summer school programs, instead providing instruction via the Internet
(Krafcik 2010; Olster 2010). Schools are also contracting with online providers to deliver
courses that they do not feel they could otherwise afford.
15. As of late 2010, online learning opportunities were made Definition of Online Learning
available to students in 48 states and Washington, D.C. “Online learning” refers to instructional
These opportunities were offered by a number of different environments supported by the Internet.
providers, including state virtual schools, multidistrict full- Online learning comprises a wide variety
of programs that use the Internet within
time online schools, single-district programs and programs and beyond school walls to provide
run by consortia or postsecondary institutions (Watson et al. access to instructional materials as well
2010). Companies in the private sector also provide online as facilitate interaction among teachers
and students. Online learning can be fully
learning opportunities for secondary students. online or blended with face-to-face
interactions. Each of these approaches is
described below.
According to survey-based estimates by the International
Association for K-12 Online Learning (iNACOL), 1.5 Fully online learning is a form of
distance education in which all instruction
million students took one or more online courses in 2010 and assessment are carried out using
(Wicks 2010). In these courses, students received all or part online, Internet-based delivery (Picciano
of their instruction over the Internet and interacted online and Seaman 2009; U.S. Department of
Education 2007). In this brief, both
with teachers, peers and digital learning content. Some teacher-led instruction and resources
states such as Alabama, Florida and Michigan have made designed to instruct without the presence
the online learning experience part of their graduation of a teacher meet the definition of fully
online learning if they include
requirements (Watson et al. 2010). instructional environments accessed
exclusively through the Internet.
Online learning has become popular because of its Blended learning (also called hybrid
perceived potential to provide more flexible access to learning) allows students to receive
significant portions of instruction through
content and instruction by both face-to-face and online means.
Researchers see blended learning in the
middle of the spectrum between fully
1) increasing the availability of learning experiences face-to-face and fully online instruction
for those who cannot or choose not to attend (Graham, Allen, and Ure 2005; U.S.
traditional schools, Department of Education 2007; Watson
et al. 2010).
2) assembling and disseminating instructional content
more efficiently, and
3) increasing student-instructor ratios while achieving learning outcomes equal to those
of traditional classroom instruction.
Some proponents see technology as having potential beyond increasing efficiency in
instructional delivery, for example, by providing a community of learners to support
understanding of a complex body of knowledge (Riel and Polin 2004; Schwen and Hara
2004). Online technologies can expand and support such communities by promoting
16. “participatory” education models rather than simply changing education delivery modes
(Barab, Squire, and Dueber 2000; Barab and Thomas 2001).
Others argue that online learning can provide individualized and differentiated instruction
(Archambault et al. 2010; Christensen and Horn 2008; Waldeck 2008; Watson and Gemin
2008) through multiple mechanisms that provide immediate formative feedback about a
student’s performance (Dennen 2005; Rice et al. 2008) or through modularized content that
enables learning the same content at a different pace or to achieve different learning goals.
The distinction between fully online and blended learning is important in part because it
helps set the standard for comparing costs and outcomes (Watson et al. 2010; U.S.
Department of Education 2010b). To be viewed as a success, online programs that provide
access to courses or programs that would otherwise be unavailable need to be as effective as
traditional alternatives. Blended approaches are typically perceived as quality improvements
that enhance and improve traditional instruction but as such need to demonstrate gains in
learning quality or rate of learning to justify the additional expenses. Although the terms
fully online and blended are commonly used and conceptually useful, blended learning itself
can take many forms, and models of blended instruction are still emerging. If either fully
online or blended instruction can transform instructional processes, there is an opportunity to
improve quality as well as reduce costs. 3
Purpose of this Report
Because online learning is serving increasing numbers of secondary students, it is essential
to understand whether, when and how particular implementations of online learning are
equally or more productive than other forms of instruction. The purpose of this report is to
support educational administrators and policymakers in becoming informed consumers of
information about online learning and its potential impact on educational productivity. The
report provides foundational knowledge needed to examine and understand the potential
productivity contributions of online learning and reviews the research that describes how
online learning might offer productivity benefits compared with traditional brick-and-mortar
The literature describes “traditional” designs as offering the online equivalent of simple didactic instruction (e.g.,
programs in which the system provides content to read and a quiz at the end), whereas “transformational” designs provide a
fundamentally different student experience. See Watson et al. [2010] for a more detailed description of transformational
practices in one state virtual school.
17. This report includes
• a framework for understanding general principles associated with systematic
productivity analyses in education;
• a summary of claims regarding how online learning could affect educational
• a review of the literature that informs understanding of the costs and effects of
online learning relative to traditional face-to-face instruction;
• a discussion of the implications of the findings; and
• brief summaries of specific resources for readers wanting to learn more about
topics addressed (appendix).
Ultimately, the hope is that this information can be used to help educators realize
productivity improvements in the future.
18. Introduction to the Measurement of
Educational Productivity
A number of publications describe formal analytic procedures for estimating the
productivity of countries (e.g., Organisation for Economic Co-operation and Development
[OECD] 2008), industries (e.g., Colecchia and Schreyer 2001) and firms (e.g., Brinkerhoff
and Dressler 1990). In education, productivity is typically understood as a ratio of the cost of
inputs per output, with outputs often measured in terms of student academic attainment
(Cohn and Geske 1990; Levin and McEwan 2001). The academic literature provides at least
three frameworks for analyzing educational productivity. Levin and McEwan (2001) present
a series of detailed frameworks for the rigorous analysis of costs and outcomes specific to
educational interventions. Rumble (1997) and Kaestner (2007) apply similar frameworks
specifically for distance learning and online learning. With rare exceptions, productivity
analyses typically require that at least one other alternative to the option under study be
examined because cost-effectiveness and similar ratios are relative.
As suggested above, there are several methodologies used to measure productivity. These
methods all measure costs and are often classified by the type of outcome measure used.
This report focuses primarily on what is known as a cost-effectiveness approach because
effectiveness data are in many cases available in education and because these outcome
measures, such as test scores, retention rates and school attendance, are considered
meaningful and consequential in educational contexts, especially in K–12. 4 Cost-benefit
approaches, which calculate monetary value for outcomes of interest in order to create a
dollars-to-dollars comparison, are often used to support decisions about government
programs because they allow comparison of projects across industries (e.g., health and
education). However, this type of study can be particularly challenging in education, both
because of the difficulties associated with assigning monetary value to many educational
outcomes (such as test scores) and because measurements of the types of outcomes that are
more readily monetized (such as impact on wages and income over time) would rely on data
not currently collected or readily available.
Cost-effectiveness studies often have two goals: (1) to inform decisions about a particular
program under way in a particular location and (2) to inform other stakeholders as they
consider undertaking new programs of their own. To accomplish these goals, studies must
Educational attainment measures are consequential in the sense that they are used to determine grade progression, college
attendance, etc.
19. both measure the productivity of the online learning intervention relative to other
alternatives and describe key elements of the intervention that will be needed for replication
of the program elsewhere. Cost-effectiveness ratios are helpful only in the context of
comparable ratios based on realistic alternatives, which in turn suggests the need for
consensus among analysts regarding the comparability of those alternatives. For example, a
cost-effectiveness study might look at the cost of a range of alternatives for raising student
achievement in math by an average of 10 points per student. When the results of the analysis
are expressed this way, they can be used to compare similar costs and outcomes across
design alternatives. Studies such as these can inform a host of decisions related both to
currently operational and to planned online learning environments. For example, educational
administrators could ask
• Would online learning help us serve more summer school students at less cost
without sacrificing outcomes?
• Over the last 3 years, has our online learning system been more or less
productive than our traditional brick-and-mortar alternatives in terms of the cost
per completer, average grades or test scores? Is this ratio likely to change over
the next three years?
• How do blended and strictly online versions of a given course compare in terms
of the cost to improve students’ learning outcomes?
Clearly, depending on the question to be answered, different sources of data will be needed.
Although the costs and outcomes for any online learning system will vary according to its
design and scope, both costs and outcomes in education generally fall within a common set
of categories that can be used to guide cost and outcome measurement (Exhibit 1).
20. Exhibit 1: Components of Educational Productivity Analyses
Note: The bullets provided in each category in this exhibit are representative rather than
Exhibit reads: Several different kinds of costs account for final cost estimates developed in
educational productivity studies.
21. Estimating Program Costs
The “ingredients” approach to specifying costs adapted from Levin and McEwan (2001)
suggests determining all the types of costs associated with developing and running a
program and assigning a value to each (see appendix for additional resources that look at
this approach more deeply). Although specific costs will vary by program, they can be
summarized using the categories in Exhibit 1.
• Personnel costs include the time of teachers, teaching assistants, developers,
administrators and any others involved in creating or running the online learning
• Facilities costs include the costs of buildings, classrooms, office space and furniture
for administrative and instructional purposes as well as for housing computers.
• Equipment and infrastructure costs include the resources required to implement
needed technology, support its operation and maintain the equipment and
infrastructure in working order.
• Materials and supplies costs include purchased online curricula or textbooks, as well
as other physical goods or processes (e.g., the costs associated with printing and
• The other category varies by implementation model but usually includes general
operational services and student supports required for the successful and legal
operation of the program or school. If productivity analyses of online and other
instructional practices are to become more routine, accounting systems at every level
of education need to better identify costs in each of these categories.
A few basic considerations for cost analyses are as follows:
• Rigorous cost analyses include the costs or value of all resources essential to an
intervention as well as its most realistic alternative, and the same types of costs
are included for each alternative so that apples are compared with apples. For
example, the initial costs associated with program planning and curriculum
development are often important considerations for new online systems.
Similarly, investments in technology such as hardware and connectivity are often
required before starting an online learning program. On the other hand, facilities
costs can be sizable for more traditional instructional approaches. In estimating
the costs of each alternative, costs of planning, curriculum development and
22. facilities should be treated similarly across alternatives. 5 Significant mismatches
across cost estimates can lead to erroneous conclusions.
• All components of a program, regardless of the source of funding, should be
included. For example, teachers’ time may be covered by their contracts and
therefore not entail an additional cost incurred by the online learning program.
However, if teachers spend time providing online instruction, the system incurs
an “opportunity cost” for other possible uses of those hours that are lost. The
same is true of shared resources such as computer labs, even if the computers
were not purchased specifically for the online learning program.
• Available time, data and budgets for cost-effectiveness research will also shape
the precision of cost estimates and the rigor of outcomes research. In most
circumstances, it would be prohibitively expensive to arrive at a fully accurate
cost, and it is often difficult to obtain accurate records of each component cost.
Accordingly, most studies include some level of estimation. The balance of
actual and estimated costs is a function of many factors, including the time and
funding available for the productivity study, the accuracy of available records,
the research team’s access to cost data that may be proprietary and the goals of
the study. Precision is more important for some purposes than for others. Specific
line items in the framework may require a combination of actual and estimated
costs; for example, it is relatively easy to identify hourly teacher salaries, but
individual teachers may not have tracked the hours they spent in developing
curriculum for an online learning program. Similarly, information about student
motivation is often considered on an ad hoc basis if at all. Questions about
systems currently being implemented should be supported by data on actual costs
and outcomes; analyses of future or planned systems will necessarily rely on
estimates based on historical spending and on costs/outcomes for similar
• In estimating costs, the same parameters should be used across conditions so
that estimates are comparable. In studies that compare an online learning system
with a traditional model, state per-pupil education expenditures are a common
proxy for the per-pupil costs of the traditional condition. However, that proxy
should be used with care, because neighborhood schools typically include a wide
Cost analysts use mathematical procedures to develop annual costs for investments that may be purchased in one year but
used across several years. For example, if a computer is purchased, the annual cost of the hardware would be based on the
initial cost divided by the estimated life of the computer (often considered three years) and may include additional
adjustments. Ongoing costs like computer maintenance should also be considered. A more detailed treatment of costing
procedures is beyond the scope of this report, but related resources are provided in the appendix for the reader interested in
learning more.
23. range of school services not normally provided through online instruction. For
example, public schools are required to provide services for students with special
needs, and they support facilities such as gymnasiums and cafeterias. A state’s
average per-pupil funding may be a good starting point for estimating the cost of
traditional schooling, but a more in-depth analysis will also need to consider the
specific costs for parallel services associated with online and brick-and-mortar
Documenting Context and Implementation
Specifying essential design components of both an intervention and its alternatives is
particularly important. Fundamental elements of educational implementation include
teachers, materials and students (Cohen and Ball 1999). Lovett, Myers and Thille (2008)
attributed their positive results with online learning in a university statistics class to design
features based on learning science research. Those features include clear organization and
structure of online materials, frequent opportunities for students to practice new knowledge
or skills, immediate and targeted feedback, and effective media use. Other fully online
implementations may realize different learning outcomes if they do not incorporate these
characteristics. Costs may also differ as a result. Similarly, blended learning programs that
use teacher time effectively are likely to produce stronger cost-effectiveness results than
blended learning programs that use the time poorly. Multiple cost-effectiveness ratios would
be needed to capture significant variation in or different models of implementation. Given
that the research base is evolving, policymakers and administrators should carefully consider
the applicability of findings from particular studies to other contexts. For example, study
findings that conclude that an implementation of online learning was effective in a specific
context should not be extrapolated to suggest that all forms of online learning are effective
for everyone. (The appendix lists reference to additional resources that provide quality
guidelines for online learning.)
Measuring Program Outcomes
Any given program is likely to have a range of possible outcomes. Selecting the most salient
outcome or outcomes to measure is a case-by-case decision. Exhibit 1 lists several types of
possible outcomes.
• Learning outcomes are most often measured at the student level, although they may
be aggregated at the classroom, school or district level. Stakeholders might be
particularly interested in scores on standardized tests; some studies also look at
24. grades or alternative assessments that are more sensitive to the types of skills to be
developed through innovative instruction. It is particularly important to consider not
just numerical outcomes (e.g., graduation rates and test scores), but also the quality
of those student outcomes.
• Affective outcomes relate to factors such as student motivation, academic
engagement and future goals. These outcomes are often measured through student
• Student school success includes grade promotion, retention and graduation. At
secondary schools, advanced course taking such as Advanced Placement (AP) course
enrollment and dual enrollment in college are also considered important student
success measures as they reduce the time and money required to graduate from
college (Greaves et al. 2010).
• Staff outcomes include retention of teachers and other staff members as well as
improving quality of instruction through professional development.
• System outcomes from an online learning program may also be desired. These might
include increases in (1) student access to instruction and qualified teachers, (2) scale
of operations and (3) rate of student learning. Qualitative improvements can also be
The rate of learning, which is based on time necessary for students to reach a given level of
mastery, is an outcome not often used in K–12 education. As the old proverb goes, “Time is
money,” and savings are possible if students need less time to meet instructional goals. A
similar issue relates to opportunities to increase students’ time on task (Cavanaugh 2009a).
Increases in time on task have been associated with improved student outcomes. Two
examples of ways to increase student time on task are homework (to increase time on task
outside the classroom) and increased engagement (so that students are thinking more deeply
in the classroom and willing to spend more of their personal time on academic tasks).
Evaluating both will be important as technology changes the types of evidence available for
documenting student learning and the ways in which that evidence can be used to improve
and inform instructional environments (U.S. Department of Education 2010a). 6
The appendix provides pointers to additional documents that include conceptual frameworks and empirical evidence
associated with online learning for the reader interested in learning more.
25. As with any type of effectiveness study, estimating outcomes for productivity need to follow
commonly accepted guidelines for program evaluation (e.g., see Frechtling and Sharp 1997).
The outcomes selected to measure and report should be the most important to the
intervention and attributable specifically to the program. For example, if the program is a
single mathematics class that students can take either face to face or online, gains in
mathematics test scores for participating students at the end of the year could be evaluated;
however, it may be difficult to draw clear connections between a specific program and larger
academic outcomes (e.g., graduation) that could be influenced by a host of factors outside
the scope of the mathematics class. A study also may take into account multiple outcomes as
is common in cost-benefit or cost-utility studies (Levin and McEwan 2001).
Cost-Effectiveness Research Requirements
The above productivity framework suggests a number of requirements for sound studies of
online learning productivity that are offered here to help guide literature analysis.
1. Specify important design components of the intervention. Because the costs and
outcomes of different programs can vary widely, simple comparison of ratios will do
little to elucidate the factors that contributed to productivity gains. For study results
to suggest design features worthy of replication, the study must describe important
variations between the treatment and control conditions. In some studies, the only
salient differences are related to technology and delivery systems. However, well-
designed online learning interventions generally include modifications in pedagogy
and curricular materials and other enhancements that take advantage of the
technology platform. These factors may not be included in a productivity ratio, but
including them in research reports is important for supporting the interpretation of
reported productivity ratios.
2. Compare at least two conditions. On its own, a ratio of per-pupil program cost per
unit of outcome is not meaningful. An online and control condition (e.g., comparing
the ratio of costs and test scores in an online academy with that in a brick-and-mortar
school teaching the same content) can be used to measure changes in productivity
ratio across conditions. Analysts should also consider blended alternatives, not just
contrasts between fully online learning and fully face-to-face instruction.
3. Measure both costs and outcomes. Both costs and outcomes can vary widely across
implementations, suggesting that both factors must be measured for each
intervention under study. The study should use the same cost framework for the two
conditions so that all relevant costs—and the same costs—are compared. Similarly,
26. any relevant outcomes can be selected as long as the same outcomes are measured
for both conditions.
4. Relate costs and outcomes using a single ratio for each model under study.
Following Levin and McEwan (2001) and other research, the framework presented
here describes productivity as a ratio of costs and outcome (e.g., cost per additional
graduate of a high school program). This ratio allows the comparison of online
learning systems with other alternatives under consideration.
5. Control or hold constant other factors not related to the online learning-supported
intervention. Factors such as student population, curriculum content, course duration
and the amount of time students spend engaged in learning can strongly affect the
outcomes in each condition. As is the case for any high-quality research, these
factors must be held constant so that the only predictable variation across conditions
is the design feature under study. This includes controlling for prior student
achievement and other important factors. For example, if students in an online
program score higher on achievement tests than their peers in traditional instruction,
a more effective learning environment online could have influenced that result;
alternatively, the online environment may simply have attracted a higher achieving
group of students or other specific student subgroup.
The measurement of educational productivity requires systematic consideration of the costs
and outcomes of an intervention compared to a range of alternatives. Analyses should also
document implementation features of the intervention so that it can be appropriately
replicated elsewhere and so that readers understand how costs were transformed into
outcomes. A review of the research conducted for this report found that the available
literature base regarding the productivity of online learning is fragmented and spotty at best.
For example, no rigorous analyses were found that illustrated the guidelines for cost-
effectiveness research described above. Studies that compared two conditions typically did
not look at parallel costs or did not control for student characteristics. 7
Two research reports—an audit for the Wisconsin State Legislature (Stuiber et al. 2010) and a study of the Florida Virtual
School (Florida Tax Watch Center for Educational Performance and Accountability 2007)—include data about costs and
effects. Both found that online learning environments produced better outcomes than face-to-face schools and at a lower
per-pupil cost than the state average. However, these conclusions must be viewed cautiously because both reports lacked
statistical controls that could have ruled out other explanations of the findings.
28. The Productivity Potential of Online Learning
The use of technology as a productivity tool has a much longer history in business than in
education. Research from industry generally suggests that information and computer
technologies can play an important role in improving productivity. These gains, however,
have typically been realized only when technology is coupled with fundamental
organizational changes that re-engineer business processes, taking advantage of the
affordances of the tools to work smarter and more efficiently (Athey and Stern 2002;
Atkinson and McKay 2007; Brynjolfsson and Hitt 2000). For example, case studies of eight
industries suggest that service industries such as hotel and retail banking made significant
investments in information technology, but initially enjoyed more modest payoffs than other
industries, in part because they often missed opportunities to use automatically generated
data about customers and purchases to inform business decision making (McKinsey Global
Institute 2000).
To understand the potential for educational productivity offered by online learning
opportunities, it is similarly necessary to look at the pedagogical and practical affordances
through which productivity gains might be realized. Online learning is often suggested as a
means for improving educational outcomes, expanding access at lower costs than
conventional approaches or allowing talented teachers to focus on what they do best by
automating or offloading more routine tasks (Christensen and Horn 2008; Christensen,
Johnson, and Horn 2008; Moe and Chubb 2009; Olster 2010; Wilson 2010; Wise and
Rothman 2010).
A review of the literature was conducted to gather empirical research that provides evidence
of actual productivity impacts when online learning is compared with place-based
instruction in secondary schools. However, the available research base was found to be
lacking because studies did not adopt rigorous methodologies or did not provide comparable
information about alternatives. Given the limitations of the research specifically regarding
the costs and effects of online instruction for secondary students, the review that follows
also draws on examples and research about the use of online learning for postsecondary
Information about practices in postsecondary education is provided as illustrations and
suggestions of principles that may help inform the development of emerging practices in
secondary education. For example some studies have shown that universities that use online
learning enjoy significant savings (Buzhardt and Semb 2005; Cohen and Nachmias 2006);
Gordon, He, and Abdous 2009; Twigg 2003a) and can increase student rates of learning
29. (Lovett, Meyer, and Thille 2008). The sidebar describes the National Center for Academic
Transformation’s efforts to use online learning as a component in its redesign of
postsecondary courses. Nonetheless, findings from studies in higher education should be
applied with caution to secondary education, as student populations, learning contexts and
financial models are quite different across these levels of schooling. 8
A review of this extended literature base suggests nine different pathways through which
online learning might contribute to improved productivity. These nine pathways are not
necessarily mutually exclusive. They are illustrated here through examples in order to
demonstrate the kinds of tools and trade-offs needed to realize productivity gains. Five
address improving educational access and effectiveness, and four relate more directly to
potential cost reductions.
Wide variations in student motivation, technological fluency and developmental stages imply that different considerations
need to be made in designing online learning content (e.g., customizability for a broader range of students including
students with learning disorders) and support structures (both behavioral and academic; e.g. instruction on how to study
independently and manage time) for online learning in higher education and K-12. For example, K–12 students in general
are likely to be less prepared than college students to undertake independent learning and thus require more support.
Differences between higher education and K–12 in teacher professional development practices, scale of operation,
availability of analytic support and funding methods may also necessitate different approaches to designing online learning
infrastructure. Decentralized operations of colleges and universities, for example, pose a particular challenge to
implementing technology-based learning environments to increase productivity in higher education (Miller 2010).
30. The National Center for Academic Transformation Course Redesign Initiatives
The National Center for Academic Transformation (NCAT) works with postsecondary institutions to
improve learning while reducing costs by redesigning large-enrollment introductory courses through
technology. NCAT’s first course redesign program yielded an average 40 percent cost reduction
among all 30 participating institutions, which in NCAT’s estimation translated to a total of $3.6 million
saved each year (Twigg 2003b).* Additionally, 22 of the 30 projects supported by the Pew Charitable
Trust-funded program showed statistically significant increases in student learning as measured by
course exams, while the other eight showed learning equivalent to that in traditional formats (Twigg
2004a). Since then, NCAT has been scaling up its course redesign efforts, with six redesign models
(see below) and 70 completed projects.
Course Redesign Models
1) Supplemental Model: Supplements the traditional course with technology-based, out of class
activities. Active learning may be also promoted in a large lecture hall setting.
2) Replacement Model: Reduces the number of in-class meetings. Some in-class time is replaced
with out-of-class, online, interactive learning activities. Significant changes may be also made to
the remaining in-class time.
3) Emporium Model: Replaces lectures with a learning resources center model featuring interactive
computer software and on-demand personalized assistance.
4) Fully Online Model: Eliminates all in-class meetings and moves all learning experiences online,
using Web-based commercial software that provides automated assessments and feedback,
multimedia resources, and alternate staff models.
5) Buffet Model: Customizes learning for each student based on background, learning preference,
and learning goals and offers an assortment of individualized paths to reach the same outcomes.
6) Linked Workshop Model: Remedial /developmental instruction by linking workshops that offer
students just-in-time supplemental academic support to core college level courses.
Source: http://www.thencat.org/PlanRes/R2R_ModCrsRed.htm
* As with any reported costs, it is important to be aware of how costs are calculated— what estimates include
and what they do not. NCAT’s costing methodology does not include development or transition costs, focusing
primarily on savings associated with the reallocation and use of instructional staff time (e.g., full-time faculty,
adjuncts, teaching assistants). NCAT also focuses on percentage changes rather than total or per-student costs.
NCAT argues that it actually underreports cost savings because its methodology also excludes savings
associated with higher course retention and lower course repetition rates (Twigg 2003c). Independent review of
the data suggests that the savings may be closer to $2.4 million (Miller 2010).
31. Increasing Educational Access and Effectiveness
Many online learning programs for secondary students have been evaluated, but little
experimental or quasi-experimental research is available regarding the effect of these
programs on student learning outcomes. A recent meta-analysis (U.S. Department of
Education 2010b) looked at rigorous research in online learning generally and found that
students tended to perform better in blended learning courses than in traditional face-to-face
classes. 9 Learning outcomes for purely online instruction were equivalent to those of purely
face-to-face instruction. The meta-analysis results also suggest that the effectiveness of
online learning is quite stable across different content and learner types. Effectiveness did
not vary significantly with learner age or content area.
However, a relatively small number of studies addressed secondary students, suggesting
caution in attempts to generalize findings to secondary school populations. Only five of the
45 studies included in the meta-analysis focused on K–12 students, and those five studies
looked exclusively at blended online learning programs. None of the K–12 studies addressed
fully online academic courses or students in fully online degree programs. Four of the five
studies found that students in the online condition performed as well as or better than their
peers in traditional courses. The exception was a study of online Spanish instruction in West
Virginia (Rockman et al. 2007), which found modest to moderate advantages for face-to-
face students compared with their online counterparts. Given this limited base, it is
particularly difficult to make statements regarding the suitability of online learning as
currently designed for a range of students with disabilities, English language learners or
others with risk factors that might discourage course completion or graduation.
In fact, online learning represents many different purposes and practices (e.g., Cavalluzzo
2004), just as face-to-face learning represents a range of practices. For example, in face-to-
face education, teachers may lecture, encourage small group activities or, most likely, adopt
a range of practices in a single course. Not surprisingly, emerging evidence indicates that
some online learning programs are more effective than others. It is also important not to
generalize fingings across programs with differing designs. As noted above and by other
researchers (e.g., Cohen and Nachmias 2006), factors such as course organization and
pedagogy can significantly affect productivity. Given that the research base is evolving,
policymakers and administrators should carefully consider how applicable the findings of
particular studies are to specific contexts. For example, study findings that conclude that an
The meta analysis included only studies of Web-based instruction with random-assignment or controlled
quasi-experimental designs (i.e., using statistical controls for possible differences between the treatment and
control groups in terms of prior achievement) that examined effects for objective measures of student learning,
discarding measures related to student or teacher perceptions of learning or course quality, student affect, etc.
32. implementation of online learning was effective in a specific context or for a specific
subgroup of students should not be extrapolated to suggest that all forms of online learning
are effective for everyone.
The following discussion and examples illustrate five ways that online learning could
increase educational productivity by improving learning opportunities:
1) Broadening access to resources and experiences;
2) Engaging students in active learning;
3) Individualizing and differentiating instruction; 10
4) Personalizing learning; and
5) Maximizing teacher and student time.
Consistent with the lack of available rigorous research on educational productivity, the
descriptions of potential productivity improvements of online learning that follow the
examples below are based on commonly espoused visions for online learning for which
more rigorous research is warranted.
1. Broadening Access to Resources and Experiences
Online learning can broaden student access to courses taught by qualified teachers in schools
that could otherwise not afford to provide these courses because of relatively small student
demand locally or the costs associated with recruiting teachers with the necessary skills and
credentials. In particular, rural schools and districts sometimes have difficulty justifying the
expense of adding teachers who would serve relatively few students. Online learning
environments might increase productivity by broadening access to certified teachers without
incurring the cost of hiring highly qualified teachers at each site. Generally speaking, this
could result in more educational opportunities being made available to a larger pool of
This report adopts the definitions of individualized, differentiated and personalized that were used in the National
Education Technology Plan (U.S. Department of Education 2010a). Individualized instruction adjusts the pace of learning
to meet the needs of individual students. Differentiated instruction draws from a variety of instructional approaches to meet
the student’s needs. Personalized learning provides content that is tailored to individual student interests (U.S. Department
of Education 2010a).
33. Example: The West Virginia Virtual School offers Spanish courses to seventh- and
eighth-grade students, using a blended model of instruction that combines face-to-face
and virtual instruction as well as paper and pencil and Web-based activities (Rockman et
al. 2007). These students attend schools in remote areas and would not otherwise have
access to Spanish instruction by certified teachers. The program was delivered by a
three-member teacher team that included a lead teacher (a certified Spanish teacher) who
was responsible for the design and delivery of the daily lesson plan and weekly phone
conversations with each class, an adjunct teacher (a certified Spanish teacher) who
provided content-related feedback by means of e-mail and voicemail and who graded
student tests and products, and a classroom facilitator (a certified teacher but not a
Spanish teacher) who guided students on site to ensure that they stayed on task and
completed assignments on time. A three-year evaluation study with a matched design at
the school level found that students in the blended condition did as well as those in the
traditional face-to-face condition on a multiple-choice test, including subtests on oral
and written comprehension of Spanish. The study also reported that the blended learning
course motivated students to continue learning Spanish in high school.
The blended Spanish course increased access to instruction provided by certified teachers,
especially those who would not otherwise have such opportunity. This was particularly
important in West Virginia, which had made two years of foreign language instruction a
requirement for all middle school students but had been experiencing a serious shortage of
licensed Spanish teachers. Additionally, the course provided a variety of opportunities for
students to be exposed to and practice Spanish. These activities included listening to Spanish
through CDs and Wimba tools and communicating in Spanish with a native speaker or
instructors during or after school hours. Moreover, the program enabled teacher teams to
build a professional community where teachers in different roles could learn from one
2. Engaging Students in Active Learning
Online learning has a potential to improve learning outcomes by replacing lecture time with
group and individual work that engages students more actively in learning, enabling greater
motivation and deeper learning (Twigg 2003a, 2003b). These activities include online
discussions, continuous assessments with immediate feedback and increased computer lab
hours where students can get one-on-one support based on the work they have done from the
online learning system, the online teacher or the face-to-face teacher. Additionally,
simulations and visualizations that make challenging abstract concepts more accessible to
students represent one demonstrated advantage of computer-based resources (see Cavanaugh
2008; Kearsley and Shneiderman 1998). Some online programs are game based, facilitating
34. situated understandings, multiple perspectives and transfer through immersive experiences
and activities (Dede 2009; Gee 2006).
Example: In Vermont, Middlebury College and K12 Inc. have recently developed
interactive language programs that provide authentic experiences for K–12 students
through immersive technologies such as 3D games and social networking (Ash
2010). For example, 3D games require students to use a foreign language in contexts
such as taking orders from customers as a waiter or dialing a virtual phone and
leaving a voice message. Embedded social networking elements allow students to
practice the language with others, including native speakers. These activities help to
deepen learning by increasing student interest and motivation and encouraging
student conceptual understanding. As a result, students may invest more time and
effort in their learning.
3. Individualizing and Differentiating Instruction
Online learning environments are often described as highly individualized and differentiated
(Archambault et al. 2010; Christensen and Horn 2008; Waldeck 2007; Waldeck 2008;
Watson and Gemin 2008). Some are designed to support the learning needs of a variety of
students such as English language learners, students with disabilities and gifted students,
while others are designed to enable flexible scheduling in order to accommodate family
travel, athletics, performances or other time-specific commitments or because a student was
hospitalized or homebound. Modularization of online course content and persistent access to
learning materials allow students to progress toward different goals or at different paces.
Effective use of multimedia, hypertext, and other design features can increase accessibility
and comprehensibility of course content for different kinds of learners—including students
with disabilities and English language learners—and help students acquire multiple
literacies (Bosseler and Massaro 2003; Morse 2003; O’Hara and Pritchard 2009; Proctor,
Dalton, and Grisham 2007; Rose and Meyer 2000). For example, mathematics and science
vocabulary can be challenging for any student, especially students learning English as a
second language. The use of hypertext to provide easy access to definitions and the use of
graphics and simulations to enhance or reinforce text descriptions can make content more
accessible to English language learners (Prichard and O’Hara 2011). Prichard and O’Hara
(2011) have found:
“… environments that support linking graphics, sound and video elements in
addition to text elements … provide students with multiple opportunities for
35. language production, task engagement and academic vocabulary development ….
Not only can various language development needs be addressed simultaneously by
promoting the use of visually engaging and language rich technologies, the ability to
use these environments encompasses many of the technology skills students need as
they graduate from high schools and work toward future careers” (p.19).
Online learning environments can also offer multiple mechanisms to provide rich feedback
and communication about student performance (Dennen 2005; Rice et al. 2008; Swan 2004).
Online assessments allow efficient data collection and analyses about individual and group
performance that would be more difficult to collect in traditional classroom environments.
By incorporating accessibility features and scaffolding, universally designed computer-
based assessments may yield more valid measurements of knowledge and skills for students
with disabilities and English language learners (Almond et al. 2010; Kopriva 2009; Rose
and Meyer 2000; Russell, Hoffmann and Higgins 2009a, 2009b). Online assessments also
allow for the collection of new kinds of information about student knowledge, skills and
abilities through embedded assessments and assessment of student performance on authentic
tasks. This rich data about student performance can inform how teachers use their time and
which instructional strategies they use for particular students. The data can also be used by
developers of online content in the service of continuous improvement efforts (e.g., Cen,
Keodinger, and Junker 2007). Immediate feedback loops established in online learning
environments can also support the customization of learning content for individual students.
Example: The Cognitive Tutor®, Web-based instructional software, provides a highly
individualized blended approach to online learning. 11 The tutoring program uses
artificial intelligence to identify weaknesses and strengths in each student’s mastery of
mathematical concepts; it adapts to each student by pacing the curriculum, selecting
problems appropriate to the student’s skill level and providing immediate feedback. This
suggests a reorganization of instruction, with student performance on computer-based
formative assessments driving instruction. Pedagogically, the Cognitive Tutor focuses on
real-world problem-solving through the use of multiple representations and tools,
including online chat and virtual whiteboards. Students use the Cognitive Tutor three
days a week to work independently and with teachers or with other students during the
remaining two days. The effectiveness of the standalone computer-based version of the
Cognitive Tutor is well documented, including a randomized controlled trial that
investigated the effectiveness of Cognitive Tutor Algebra I® with more than 400 ninth-
graders in Oklahoma (Ritter, Anderson, Koedinger & Corbett 2007). Cen, Koedinger
New York City’s School of One provides another model of individualize and differentiated learning, but Cognitive Tutor
was selected as the primary example because it has well-documented gains.
36. and Junker (2007) found that students using the tutor could reach the same level of
performance in 12 percent less time than their peers who did not use it. 12
This type of online learning might improve productivity by using instructional resources
such as books and computer-based materials as well as teachers and peers more effectively.
Productivity gains can results from focusing on specific student needs in order to improve
learning or from using student time more effectively, as students are not constrained by the
collective pace of the class. In addition, flexible scheduling and other forms of
individualization might help retain students who are otherwise at risk of dropping out of
school (Repetto et al. 2010), offering the wide range of individual and social gains that have
been well documented as outcomes of high school completion (Levin and Belfield 2009).
4. Personalizing Learning
Personalized learning draws on individual students’ specific interests. Using the definition
provided in the National Education Technology Plan (U.S. Department of Education 2010a),
personalized learning not only encompasses the individualization and differentiation
described above, but also allows students to draw on their personal interests to direct
learning objectives and content. Personalized learning can tap students’ innate curiosity and
help them deepen their learning.
Example: The Ohio State University (OSU) introductory statistics course, enrolling over
2,500 students, was redesigned based on what is called the “buffet model” (Twigg
2003a). As the name suggests, students are offered options for types of lectures and labs
they receive (e.g., large group lecture, small group problem-solving, online individual
work) based on their learning styles as measured by an online pre-course questionnaire
(Acker et al. 2003). In fall 2002, the course was delivered with three customized tracks
and demonstrated that learning gains could be achieved while reducing costs (National
Center for Academic Transformation 2003a; Twigg 2004b). Students in the redesigned
course had greater success on common exams than daytime students in the traditional
course and about the same scores as students in the evening class, which had smaller
class sizes and older students and had previously outperformed the daytime class.
Additionally, failures were reduced from 7 percent to 3 percent, withdrawals from 11
percent to 8 percent and incompletes from 2 percent to 1 percent, resulting in 248 more
Cognitive Tutor now offers a Web-based delivery option, especially for higher education institutions, going beyond
stand-alone software delivery model (http://www.carnegielearning.com/higher-ed-curricula/implementations). The majority
of research on the product thus far, including those studies cited here, has been on the stand-alone software rather than the
Web version.
37. students successfully completing the course compared with the traditional course
(National Center for Academic Transformation 2003a). In addition to improving
outcomes, cost savings were reported by replacing individually held office hours with a
help room where students can work collaboratively on difficult problems and concepts.
The help room is open for all students taking any statistics course and is staffed with
teaching assistants, adjuncts, and full-time faculty throughout the day. By making this
change, OSU saved one additional teaching assistant position. Additionally, OSU found
that students prefer Web-based problem-solving sessions to in-class sessions, leading to
a reduction of the in-class session from five times to three times a week.
5. Maximizing Teacher and Student Time
There are at least two uses of online learning to improve the use of teacher and student time.
Visualizations of learning progressions and student development made possible through
learning management systems and other online data systems may offer an opportunity to
make the educator’s workload lighter by providing targeted input to lesson planning and
attempting to address individual student needs. Students may also benefit from reflecting on
their learning progress. In addition, some online learning models are designed to transfer
certain routine activities, such as skills practice and test preparation, from teacher-based
whole- or small-group instruction to activities that students can conduct independently on a
computer. Proponents of these models claim that this use of online learning allows class
time to focus on activities and discussions that take greater advantage of teacher skills and
real-time interaction with students. The National Center for Academic Transformation has
reported replacement of routine in-class activities with online activities is particularly
notable in foreign language instruction in postsecondary settings. 13 In the redesigned
courses, grammar instruction, practice exercises, testing, writing and small-group activities
are typically moved to the online environment. This not only reduces in-class meeting time,
but also frees teachers to focus on complex activities that require face-to-face interactions
such as developing and practicing oral communication skills during the in-class time (Twigg
Example: To graduate from high school, students in New York must pass five Regents
exams. The NYC iSchool—a small high school that opened in fall 2008—uses online
learning to prepare students for the Regents exams, as well as to allow practice of basic
skills. The self-paced online test-prep courses are primarily accessed during a scheduled
class period. The online course does not require an online teacher; rather, the teacher is
Again, examples from postsecondary education can suggest possible opportunities for efficiency in secondary education,
but gains should not be assumed to transfer across contexts.
38. present in the classroom and students can seek additional face-to-face help during office
hours. This enables students and teachers to spend more classroom time on learning
projects and on solving real-world problems. Since the school is still relatively new,
published research is limited. Journalistic accounts suggest that iSchool student
attendance is higher than the city average and limited seats are in high demand
(Stansbury 2009). Productivity gains might come from reducing teacher time required
for routine tasks and reallocating it for higher value activities that require teacher
expertise such as spending personal time with students and providing detailed and
complex feedback that cannot be automated by technology.
Opportunities to Reduce Educational Costs Through Online
The previous section described potential gains in educational effectiveness that may result
from the use of online learning. This section focuses on potential cost savings in particular.
Studies that examine the costs of online learning programs in comparison with face-to-face
instruction have consistently found savings associated with online learning, although the
costs of both forms of education vary considerably. Exhibit 2 provides estimates of per-pupil
spending for online and place-based secondary instruction from six different studies. These
studies used different methods to calculate per-pupil spending. They also drew on different
data sources. Some of the data were self-reported by school leaders (Cavanaugh 2009b) or
estimated (Watson 2004). In other cases, data were verified by an external auditor (Stuiber
et al. 2010). It is important to note that none of these studies appeared in a peer-reviewed
39. Exhibit 2: Comparison of Per-Pupil Spending
A recent report estimated the average per pupil costs of various models of online learning (Battaglino, T.M., M. Haldeman,
and E. Laurans (2012). The Costs of Online Learning. Washington, DC: The Thomas B. Fordham Institute.) and found that
virtual schools are likely to cost less than blended models. Based on expert opinion, the report found that the average per
pupil cost of virtual schools ranged from $5,100 and $7,700, and the average per pupil cost of blended school models cost
between $7,600 and $10,200.
Exhibit reads: According to Stuiber and colleagues (2010), the average per-pupil spending for face-
to-face education in Wisconsin was $11,397, compared with only $6,077 per pupil for online
The idea that online learning can reduce costs has None of the studies in Exhibit 2
intuitive appeal (Christensen and Horn 2008; compared the actual costs of
Christensen, Johnson, and Horn 2008). Some cost both development and delivery
studies have found online learning to be less of parallel educational services
expensive, and it is noteworthy that this collection of for online and face-to-face
studies consistently found cost savings associated with
online learning. However, it is important for readers of
these studies to be aware that most did not include
rigorous control of the costs and program scope being
compared. For example, they often compared the costs of online learning programs (either
actual costs or allocated funding) with per-pupil allocations in the state, which cover the
broader array of services provided by traditional public schools, e.g., cafeterias,
transportation. None of the studies in Exhibit 2 compared actual costs of both development
and delivery of parallel educational services for online and face-to-face instruction.
40. Nevertheless, literature about online learning more generally suggest at least four ways that
online learning might contribute to cost reduction:
1) Increasing the rate of student learning by increasing motivation and student time
on task and helping students grasp concepts and demonstrate competency more
2) Reducing salary costs by redesigning processes to allow for more effective use of
teacher time, increasing teacher-student ratios or transferring some educational
activities to computers;
3) Reducing facilities costs by leveraging home and community spaces in addition
to traditional school buildings;
4) Realizing economies of scale by leveraging initial development costs as broadly
as possible.
1. Increasing the Rate of Learning
This is a potentially powerful pathway because it could
change the value of student time (Watson et al. 2010). Lovett, Myer and Thille (2008)
Traditionally, course credit and other measures of found that college students
educational attainment have been tied to seat time—that is, learned statistics online about
how many hours the student spends in a classroom with a 50 percent more quickly than
certified teacher. It does not matter if the student could
did students in traditional large
master the required material more quickly or not. Although
lecture courses. Fletcher and
time is money, the value of student time and its effective
use is often ignored in decisions about elementary and Chatham (2009) found
secondary education. A focus on identifying what students efficiency gains of about 30
already know and what they need to know and tailoring percent for a variety of types of
instruction based on this information to meet student needs online training for the military.
may make instruction more efficient, allowing students to
learn content more quickly or more deeply. For example,
Lovett, Myer and Thille (2008) found that college students
learned statistics online about 50 percent more quickly than students in traditional large
lecture courses. In looking primary at studies of learning in the military, Fletcher and
Chatham (2009) found efficiency gains of about 30 percent for a variety of types of online
training for the military in a study of:
41. “several hundred studies comparing standard classroom instruction (e.g., lectures,
text- and work-books, and some hands-on laboratory experience) with the use of
technology-based instruction (Fletcher 1997, 2004, 2009). Research suggests that
this finding results from the capability of computers (i.e., those who program
computers) to tailor the pace, content, and sequence of instruction to the needs of
each learner. Absent computer technology, such a capability has long been viewed as
desirable, but unaffordable” (pp. 19-20).
No similar studies involving elementary or secondary students could be located at the time
of this review. Research has not yet demonstrated whether similar learning gains per unit of
time can be realized for younger students.
2. Reducing Total Salary Costs
The Southern Regional Education Board (2006) found that the largest cost component for
both traditional brick-and-mortar schools and state virtual schools was personnel, with
typical school expenditures ranging from 70 to 80 percent of operating budget and possibly
higher for virtual schools. But cost estimates vary widely depending on the type of online
learning program (Watson et al. 2009).
Some have pointed to online learning’s potential for increasing the number of students
served by a teacher (Moe and Chubb 2009). For example, the cost savings reported by the
National Center for Academic Transformation’s course redesign work, as mentioned earlier,
were primarily derived from decreased time spent by instructional personnel (e.g., faculty,
graduate students) and substitution of less expensive personnel (Twigg 2003b). Analysts
have also suggested that some teacher tasks can be handled by technology, as is the case
when online learning is used for supplementary purposes in classrooms. Teachers in fully
online programs also suggest that more of their time is spent directly on instruction rather
than ancillary duties of school-based teachers such as classroom management and hall and
lunch duty. Focusing teacher time on instruction is one way teachers may serve more
students and ultimately reduce the number of teachers needed to maintain comparable
student outcomes (Moe and Chubb 2009). Although such a change might seem to threaten
teachers’ job security, increasing student-teacher ratios are one way of compensating for the
teacher shortages projected in the near future (Committee on Science, Engineering, and
Public Policy 2005; Ingersoll 2000; Murphy, DeArmond, and Guin 2003). 14
The National Educational Technology Plan (2010a) describes a future in which there are many differentiated roles for
teachers, suggesting the need for more educators rather than fewer, even if there is an increase in student-teacher ratios
related directly to instruction.
42. Although somewhat controversial, some higher education programs are reported to have
successfully reduced personnel costs without needing to cut full-time positions (Twigg
2003a, 2003b). These substitutions include technology-assisted instructional activities (e.g.,
online quizzes with automated scoring instead of hand grading) and lower priced labor such
as undergraduate learning assistants or newly appointed positions such as course assistants,
preceptors, and course coordinators who assume specific roles within the course, enabling
faculty to concentrate on tasks that require high-level expertise and experience. In practical
terms, these cost reduction strategies often translated into reducing the number of sections
and face-to-face meetings with full-time faculty as well as increasing the number of students
served per section or instructor.
For example, Florida Gulf Coast University (FGCU) redesigned a required course called
Understanding the Visual and Performing Arts into a fully online course. The purpose of the
redesign was to address the challenges caused by rapid enrolment growth, including
difficulty in finding qualified instructors and inconsistency in how the course was taught by
part-time instructors (Wohlpart et al. 2006). The redesign process reportedly led to a
reduction in the number of class sections from 31 to two while increasing enrollments from
800 to 1,200 (National Center for Academic Transformation 2002b). In the redesigned
course, several preceptors with undergraduate degrees in English served as teaching
assistants and were responsible for monitoring small group online discussions. Each
preceptor supported up to 60 students (10 teams of six each), which was twice the number of
students served by an instructor in the traditional model (30 students per section). The
redesigned course curriculum became more coherent and consistent through the use of a
common syllabus, textbook, set of assignments and course Web site (National Center for
Academic Transformation 2003b; Wohlpart et al. 2006). Technology was used to offload
labor-intensive activities, such as presenting course content and grading exams and papers.
The reported cost per student was reduced from $132 in the traditional format (enrolling 800
students) to $81 in the first year of redesign implementation while enrollment was increased
to 950. In the second year of implementation, the cost per student was reduced further to $70
while enrollment was increased to 1,200. FGCU reported that that this was achieved without
compromising instructional quality. Savings were realized by both increased student-staff
ratios and lower salaries for staff with lesser credentials. Comparisons of student learning in
the redesigned course and the traditional course showed that students performed better than
those from the traditional course on common tests of content knowledge and critical
thinking skills (National Center for Academic Transformation 2003b).
43. 3. Reducing Facilities Costs
Compared with traditional brick-and-mortar education, online learning can reduce the need
for physical space (e.g., when students take courses at home). Although online schools
typically have few or no costs associated with physical infrastructure (e.g., instructional
facilities, student transportation, meals), they have higher costs for technology and
instructional development. (Technology infrastructure costs may decrease with emerging
information technology solutions such as cloud computing, but development and
management costs constitute nontrivial expenses that are expected to persist.) In addition to
the costs of hardware, software, program development and maintenance/support for central
technology services, state and district programs must ensure that all students have equitable
access to the hardware and software needed to participate (Anderson et al. 2006).
Physical space costs are an important cost driver of traditional schooling. By substituting
classroom instruction with online instruction, the need for physical space can be reduced. In
fact, the University of Central Florida that implemented course redesign with the National
Center for Academic Transformation reported cost savings from delivering portions of
American government course online, reducing the amount of physical space required for the
course (Twigg 2003a). 15,16
4. Realizing Economies of Scale
A few studies in postsecondary education have found online learning to be an expensive
alternative because of the initial development costs and the personnel costs for delivering
instruction, especially when the online learning program is designed to equal or exceed the
quality of face-to-face instruction (Jones 2001; Ramage 2005; Smith and Mitry 2008).
Establishing an infrastructure that can support scale can incur significant costs as well.
To achieve overall productivity gains in these situations, it is important that some of the
financial investments associated with online learning are leveraged across many students by
reusing digital course materials. Once an online course is developed, digital resources can be
reused at a relatively low marginal cost, the term economists use to refer to the change in
total cost when the quantity produced changes by one unit—in this case, the cost of adding
Additional empirical analyses are required to understand better the trade-offs associated with reductions in facilities costs
and costs associated with the research and development of online programs, particularly in institutions creating home-
grown content.
Although this is not online learning per se, a recent estimate shows that by providing a laptop computer to each student,
schools across the country can potentially save $825 million (or $15 saving per student per year) in physical space costs
because fewer dedicated computer labs and physical space at the back of the regular classroom would be necessary in a
one-to-one mobile computing model (Greaves et al. 2010).
44. the next student to a program. Similarly, costs of Internet-based distribution seem to be
relatively low in settings where an adequate technical infrastructure is already installed.
Although the continuous improvement of courses will require some curator costs over time
to make sure materials are relevant and dynamic, these costs may be minor relative to
publishing new bound editions of books, especially when distribution costs are included.
Moreover, by conducting a bulk of learning activities online, costs associated with copying
materials (e.g., paper, ink, teacher time) and paperwork can be greatly reduced. According to
one estimate, for copying materials alone, online learning can potentially achieve a saving of
$2.2 billion per year at the national level, based on an estimate that schools save $40 per
student each year (Greaves et al. 2010).
Scale is important in any study of educational productivity and no less so for online
learning. Compared with conventional instruction, online learning may incur higher start-up
costs associated with developing a new program and perhaps for developing curriculum and
digital resources. Although online course content can be expensive to develop, once created
it has the potential to be distributed to large numbers of students (e.g., Adsit 2003;
Christensen et al. 2008; Watson 2004). However, course development may constitute only a
small portion of total costs depending on the instructional model (Anderson et al. 2006). For
example, an online course that requires teachers to replicate traditional lecture formats and
deliver the bulk of instructional content verbally to passive listeners at the same teacher-
student ratios—but does so online—will incur ongoing costs per student that may exceed the
cost of instructional materials per se. Additionally, critics of the economics of scale logic
assert that large-scale delivery of courses would reduce student opportunities for social and
affective experiences that are particularly important for developing soft skills (e.g., Bauman
46. Determining whether online learning is more or less cost-effective than other alternatives
does not lend itself to a simple yes or no answer. Educational stakeholders at every level
need rigorous information regarding effective instructional strategies and methods for
improving educational productivity. Comparison of relative productivity requires attention
to a host of factors, including the students served, the subject domain, scale, budget and
design factors such as the role of the teacher and the level of blending of online and face-to-
face components (Liu and Cavanaugh in press). The framework presented in this report is
intended to serve as a resource for stakeholders’ use in assessing available research. It can
also be used as a guide for designing future research. As the spread of online learning
alternatives continues, educational stakeholders will need to invest in productivity research
that includes comparisons of the cost-effectiveness of instructional alternatives and builds
knowledge of how best to define and measure student, teacher and system outcomes.
Studies designed to inform educational decisions should follow rigorous methodologies that
account for a full range of costs, describe key implementation characteristics and use
reliable estimates of student learning. Unfortunately, no studies were found in this review
that rigorously analyzed the productivity of online learning for elementary and secondary
students, although the available evidence suggests that online learning might improve
educational productivity if properly deployed.
With respect to costs specifically, institutions need to consider both total costs and per-
student costs of online learning relative to conventional instruction. Relatively high total
costs may be more palatable if courses can be leveraged across a wider student audience.
Moreover, cost drivers in an online environment differ from those in face-to-face
environments, suggesting a crossover point for student enrollment numbers at which one
format becomes more cost-effective than the other. Institutions must also consider whether