Meta-Analysis of Distance Education

The Effects of Distance Education on
K–12 Student Outcomes:
A Meta-Analysis
October 2004
Meta-Analysis of Distance Education
Learning Point Associates 2
The Effects of Distance Education on
K–12 Student Outcomes:
A Meta-Analysis
October 2004
Cathy Cavanaugh
University of North Florida
Kathy Jo Gillan
Duval County Public Schools
Jeff Kromrey
University of South Florida
Melinda Hess
University of South Florida
Robert Blomeyer
North Central Regional Educational Laboratory
1120 East Diehl Road, Suite 200
Naperville, Illinois 60563-1486
(800) 356-2735  (630) 649-6500
www.learningpt.org
Copyright © 2004 Learning Point Associates, sponsored under government contract number ED-01-CO-0011.
All rights reserved.
This work was originally produced in whole or in part by the North Central Regional Educational Laboratory
with funds from the Institute of Education Sciences (IES), U.S. Department of Education, under contract number
ED-01-CO-0011. The content does not necessarily reflect the position or policy of IES or the Department of
Education, nor does mention or visual representation of trade names, commercial products, or organizations
imply endorsement by the federal government.
Meta-Analysis of Distance Education
Learning Point Associates 3
Contents
Abstract ........................................................................................................................4
Introduction ..................................................................................................................5
Distance Education in the K–12 Context .............................................................5
Characteristics for Success ..................................................................................6
Teaching and Learning Theory............................................................................7
Purpose of the Study.....................................................................................................8
Method .......................................................................................................................10
Location and Selection of Studies......................................................................11
Limitations of the Review.................................................................................13
Coding of Study Features ..................................................................................13
Calculation of Effect Sizes ................................................................................14
Statistical Analysis of Effect Sizes ....................................................................15
Results........................................................................................................................15
Characteristics of the Study...............................................................................15
Overall Effects on K–12 Distance Education.....................................................16
Publication and Methodological Variables.........................................................18
Distance Education Variables............................................................................18
Instructional and Program Variables..................................................................19
Discussion ..................................................................................................................19
Implications for Research and Practice..............................................................19
Conclusions ................................................................................................................21
The Need for Prospective Study in Virtual Schooling........................................22
Recommendations for K–12 Online Learning Policy and Practice ....................23
References ..................................................................................................................26
Appendix: Coded Variables and Study Features in the Codebook ...............................32
Meta-Analysis of Distance Education
Learning Point Associates 4
Abstract
The community of K–12 education has seen explosive growth over the last decade in distance
learning programs, defined as learning experiences in which students and instructors are
separated by space and/or time. While elementary and secondary students have learned through
the use of electronic distance learning systems since the 1930s, the development of online
distance learning schools is a relatively new phenomenon. Online virtual schools may be ideally
suited to meet the needs of stakeholders calling for school choice, high school reform, and
workforce preparation in 21st century skills. The growth in the numbers of students learning
online and the importance of online learning as a solution to educational challenges has increased
the need to study more closely the factors that affect student learning in virtual schooling
environments. This meta-analysis is a statistical review of 116 effect sizes from 14 webdelivered
K–12 distance education programs studied between 1999 and 2004. The analysis
shows that distance education can have the same effect on measures of student academic
achievement when compared to traditional instruction. The study-weighted mean effect size
across all outcomes was -0.028 with a 95 percent confidence interval from 0.060 to -0.116,
indicating no significant difference in performance between students who participated in online
programs and those who were taught in face-to-face classrooms. No factors were found to be
related to significant positive or negative effects. The factors that were tested included academic
content area, grade level of the students, role of the distance learning program, role of the
instructor, length of the program, type of school, frequency of the distance learning experience,
pacing of instruction, timing of instruction, instructor preparation and experience in distance
education, and the setting of the students.
Meta-Analysis of Distance Education
Learning Point Associates 5
Introduction
The community of K–12 education has seen explosive growth over the last decade in distance
learning programs, defined as learning experiences in which students and instructors are
separated by space and/or time. While elementary and secondary students have learned through
the use of electronic distance learning systems since the 1930s, the development of online
distance learning schools is a relatively new phenomenon. Online virtual schools may be ideally
suited to meet the needs of stakeholders calling for school choice, high school reform, and
workforce preparation in 21st century skills. The growth in the numbers of students learning
online and the importance of online learning as a solution to educational challenges has increased
the need to study more closely the factors that effect student learning in virtual schooling
environments.
Beginning in the 1930s, radio was used simultaneously to bring courses to school students and to
help teachers learn progressive Deweyan methods of teaching (Bianchi, 2002), in what might
have been among the earliest professional development school models. From that point on,
television, audio and videoconferencing, the Internet, and other technologies have been adapted
for the needs of young learners. This meta-analysis is a statistical review of web-delivered K–12
distance education programs between 1999 and 2004 conducted in order to determine how
student learning in online programs compares to learning in classroom-based programs, and to
identify the specific factors that influence student learning.
Distance Education in the K–12 Context
The many thousands of K–12 students who participate in online education programs are attracted
to virtual schooling because it offers advantages over classroom-based programs. Among the
benefits of distance education for school-age children are increases in enrollment or time in
school as education programs reach underserved regions, broader educational opportunity for
students who are unable to attend traditional schools, access to resources and instructors not
locally available, and increases in student-teacher communication. Students in virtual schools
showed greater improvement that their conventional school counterparts in critical thinking,
researching, using computers, learning independently, problem-solving, creative thinking,
decision-making, and time management (Barker & Wendel, 2001). Academic advantages over
traditional classroom instruction were demonstrated by students in Mexico’s Telesecundaria
program, who were “substantially more likely than other groups to pass a final 9th grade
examination” administered by the state (Calderoni, 1998, p. 6); by students taking a chemistry by
satellite course (Dees, 1994); and by students learning reading and math via interactive radio
instruction (Yasin & Luberisse, 1998). Virtual school developers and instructors continue to
refine their practice, and in so doing, they learn from reports of both successful and unsuccessful
programs.
Virtual schooling, like classroom schooling, has had limited success in some situations. In an
online environment, students may feel isolated, parents may have concerns about children’s
social development, students with language difficulties may experience a disadvantage in a textheavy
online environment, and subjects requiring physical demonstrations of skill such as music,
physical education, or foreign language may not be practical in a technology-mediated setting.
Meta-Analysis of Distance Education
Learning Point Associates 6
For example, Bond (2002) found that distance between tutor and learner in an online
instrumental music program has negative effects on performance quality, student engagement,
and development and refinement of skills and knowledge. While distance learning was viewed as
beneficial for providing the opportunity for elementary school students to learn a foreign
language, Conzemius and Sandrock (2003) report that “the optimal learning situation still
involves the physical presence of a teacher” (p. 47). Virtual school students show less
improvement than those in conventional schools in listening and speaking skills (Barker &
Wendel, 2001). Highly technical subjects such as mathematics and science have also proven to
be difficult to teach well online. The Alberta Online Consortium evaluated student performance
on end-of-year exams among virtual school students across the province, and found that virtual
school student scores in mathematics at grades 3, 6, 9, and 12, and the sciences at grades 6 and 9
lagged significantly behind scores of nonvirtual school students (Schollie, 2001).
Given instruction of equal quality, groups of students learning online generally achieve at levels
equal to their peers in classrooms (Kearsley, 2000). Equality between the delivery systems has
been well documented over decades for adult learners, and while much less research exists
focusing on K–12 learners, the results tend to agree. “Evidence to date convincingly
demonstrates that , when used appropriately, electronically delivered education—‘e-learning’—
can improve how students learn, can improve what students learn, and can deliver high-quality
learning opportunities to all children” (National Association of State Boards of Education, 2001,
p. 4). Many studies report no significant differences between K–12 distance education and
traditional education in academic achievement (Falck et al, 1997; Goc Karp & Woods, 2003;
Hinnant; 1994; Jordan, 2002; Kozma et al, 2000; Mills, 2002; Ryan, 1996), frequency of
communication between students and teachers (Kozma et al), and attitude toward courses
(McGreal, 1994).
Although various forms of technology-enabled distance education for pre-college students have
been in use for nearly a century, rapid change in technology and the educational context have
resulted in a small body of research relevant to today’s conditions that can serve to guide
instructors, planners, or developers. The temptation may be to attempt to apply or adapt findings
from studies of K–12 classroom learning or adult distance learning, but K–12 distance education
is fundamentally unique.
Characteristics for Success
A primary characteristic that sets successful distance learners apart from their classroom-based
counterparts is their autonomy (Keegan, 1996) and greater student responsibility (Wedemeyer,
1981). By the time they reach higher education, most adults have acquired a degree of autonomy
in learning, but younger students need to be scaffolded as part of the distance education
experience. Virtual school teachers must be adept at helping children acquire the skills of
autonomous learning, including self-regulation. Adult learners more closely approach expertise
in the subjects they study and in knowing how to learn, due to their long experience with the
concepts and with meta-cognition, whereas children are relative novices. This distinction is
important because experts organize and interpret information very differently from novices, and
these differences affect learners’ abilities to remember and solve problems (Bransford, Brown, &
Meta-Analysis of Distance Education
Learning Point Associates 7
Cocking, 1999), and their ability to learn independently. Expert learners have better developed
metacognition, a characteristic that children develop gradually.
A second characteristic that differentiates successful distance learners from unsuccessful ones is
an internal locus of control, leading them to persist in the educational endeavor (Rotter, 1989).
Research has found that older children have more internal locus of control than younger children
(Gershaw, 1989), reinforcing the need for careful design and teaching of distance education at
K–12 levels. Younger students will need more supervision, fewer and simpler instructions, and a
more extensive reinforcement system than older students. Effective online programs for young
learners include frequent teacher contact with students and parents, lessons divided into short
segments, mastery sequences so student progress can grow in stages, and rewards for learning
such as multimedia praise and printable stickers or certificates.
Young students are different from adult learners in other ways. Piaget’s stages of cognitive
development, in particular preoperational (2 to 7 years), concrete operational (7 to 11 years), and
formal operational (11 years to adulthood) outline the phases in development toward adulthood.
The stages offer pedagogical guidance for delivering effective web based education, which
should focus on the major accomplishments of learners in these stages. Each stage is
characterized by the emergence of new abilities and ways of processing information (Slavin,
2003, p. 30), which necessitates specialized instructional approaches and attention to each child’s
development. Since adults have progressed through these stages of cognitive development,
delivery of web based education at the adult level need not concentrate on methods that help the
learner develop these cognitive skills. In contrast, web-based instruction for students in their
formative years must include age appropriate developmental activities, building on the students’
accomplishments in and through the cognitive stages. For example, an online mathematics or
science lesson designed for students at the preoperational stage needs to use very concrete
methods, such as instructing the student to develop concepts by manipulating and practicing with
real-world objects. The concept can built upon for students in the concrete operational stage
using multimedia drag-and-drop manipulations and representations, or realistic simulations. At
the formal operational stage, students are capable of using symbols, language, and graphic
organizers to continue to learn the concepts in more abstract ways.
Teaching and Learning Theory
Piaget helps us to understand that learning should be holistic, authentic, and realistic. Less
emphasis should be placed on isolated skills aimed at teaching individual concepts. Students are
more likely to learn skills while engaged in authentic, meaningful activities. Authentic activities
are inherently interesting and meaningful to the student. Web-based technology offers a vast
array of opportunities to help expand the conceptual and experiential background of the student
(Bolton, 2002, p. 5).
Neo-Piagetian theorists have expanded on Piaget’s model of cognitive development. Among
others, Vygotsky proposed that historical and cultural context play significant roles in helping
people think, communicate, and solve problems, proposing that cognitive development is
strongly linked to input from others. Vygotsky’s theory implies that cognitive development and
the ability to use thought to control our own actions require first mastering cultural
Meta-Analysis of Distance Education
Learning Point Associates 8
communication systems and then learning to use these systems to regulate our own thought
process. He believed that learning takes place when children are working within their zone of
proximal development. Tasks within the zone of proximal development are ones that children
cannot yet do alone but could do with the assistance of more competent peers or adults (Slavin,
2003, p. 43–44). When working with children using web-based technology, teachers must offer
students activities that make use of the web’s powerful tools for collaborative learning, and are
within their zone of proximal development. Online communities can provide a supportive
context that makes new kinds of learning experiences possible (Bruckman, 1998, p. 84–85).
Constructivism, a widely used theory in distance education, is founded on the premises that by
reflecting on our experiences and participating in social-dialogical process (Duffy & Cunnigham,
1996), we construct our understanding of the world we live in. Each of us generates our own
"rules" and "mental models," which we use to make sense of our experiences. Learning,
therefore, is simply the process of adjusting our mental models to accommodate new experiences
(Brooks & Brooks, 1993). Children have not had the experiences that adults have had to help
them construct understanding. Therefore, children construct an understanding of the world
around them that lacks the rich experiences that adults have had. Scaffolding or mediated
learning is important in helping children achieve these cognitive understandings (Slavin, 2003, p.
259), and are essential components of web-based learning experiences for children. Online
learning environments, when designed to fully use the many tools of communication that are
available, is often a more active, constructive, and cooperative experience than classroom
learning. In addition, technologies that are easily employed in online environments, such as mind
mapping tools and simulations, are effective means for helping students make meaning of
abstract phenomena and strengthen their meta-cognitive abilities (Duffy & Jonassen, 1992).
Purpose of the Study
With the emphasis on scientifically-based research and the call for evidence-based program
decisions in the federal No Child Left Behind Act of 2001, scientific evidence is needed to guide
the growing numbers of online school developers and educators. Many studies of K–12 distance
education have been published, but a small proportion of them are controlled, systematic,
empirical comparisons that fit the definition of “scientific,” as it is defined by the U.S
Department of Education and described at the What Works Clearinghouse website,
http://www.w-w-c.org/. This study is an effort to search for and collect the studies that fit the
definition of scientific research on K–12 distance education programs, and to draw conclusions
about the effectiveness of distance education for K–12 students based on the synthesized
findings of the studies.
Meta-analysis is an established technique for synthesizing research findings to enable both a
broader basis for understanding a phenomenon and a parsing of influences on the phenomenon.
Several recent meta-analyses related to distance education have been published in recent years
(see Table 1).
Meta-Analysis of Distance Education
Learning Point Associates 9
Table 1
Summary of recent meta-analyses in distance education
Author(s), Date Focus N of studies Effect Size
Machtmes & Asher, 2000. Adult telecourses 30 -0.0093
Cavanaugh, 2001. Academic
achievement of K–
12 students
19 +0.015
Allen, Bourhis, Burrell, &
Mabry, 2002.
Student satisfaction
among adult
learners
25 +0.031
Bernard, Abrami, Lou,
Borokhovski, Wade,
Wozney, Wallet, Fiset, &
Huang, 2003.
Student
achievement,
attitude, retention
232 +0.0128
Shachar & Neumann, 2003. Student
achievement
86 +0.37
Ungerleider & Burns, 2003. Networked and
online learning
12 for achievement
4 for satisfaction
0 for achievement
-0.509 for
satisfaction
The Sachar and Neumann study was the only one to have found a moderate effect for distance
education. Only one of the recent meta-analyses in distance education focused on K–12 learners,
and it included web-based programs along with the analog conference and broadcast programs
no longer in common use in today’s virtual schools. The explosion in virtual schools, especially
virtual charter schools in the United States, has necessitated a fresh look at the knowledge base.
The need is for research that guides practitioners in refining practice so the most effective
methods are used. Given sufficient quantity and detail in the data, meta-analysis is capable of not
only comparing the effectiveness of distance education programs to classroom-based programs,
but it can compare features of various distance education programs to learn what works. For
example, synchronous programs can be compared to nonsynchronous programs. Meta-analysis is
a tool that allows looking in detail at virtual schooling practice and results, and it can lead to
better informed practice and improved results.
Several advantages can result from a synthesis of studies of the effectiveness of distance
education programs for K–12 learners. Because all of the studies included in this review drew
data from school-based classes, the review can provide valuable insight into the practical
effectiveness of K–12 distance education. Controlled experimental research may offer findings
of theoretical interest but may not be generalizable to complex learning settings such as virtual
schools or classes. The uncontrollable cultural and social variables naturally present in a school
or class, whether online or on-ground, make a statistical synthesis a more exact test of the
strength of K–12 distance education. The effects of virtual learning would have to be strong and
consistent to be measurable across a range of natural milieus.
The purpose of this meta-analysis is to provide a quantitative synthesis of the research literature
of web-based K–12 distance education from 1999 to the present, across content areas, grade
Meta-Analysis of Distance Education
Learning Point Associates 10
levels, and outcome measures. The first goal was to determine the effects of distance education
on K–12 student outcomes, specifically academic achievement. The second goal was to identify
the effects on student outcomes of the features of distance education, including content area,
duration of use, frequency of use, grade level of students, role of the instructor, type of school,
timing of interactions, and pacing of the learning.
From the literature, the meta-analysis seeks to answer the following questions:
1. Is distance education as effective, in terms of student achievement, as classroom-based
instruction?
2. To what extent are student outcomes related to the features of a distance education system
(duration of use, frequency of use, role of the instructor, timing of interactions, and pacing of the
learning)?
3. To what extent are student outcomes related to features of the educational context (content
area, school type, and grade level)?
4. To what extent are results related to study features (year, type of publication, various potential
threats to validity)?
Meta-analysis, the use of statistical analysis to synthesize a body of literature, is appropriate for
answering questions such as these because it allows comparison of different studies by
computing an effect size for each study. Meta-analysis is used to estimate the size of a
treatment’s effect, and allows investigation into relationships among study features and outcomes
(Bangert-Drowns, 2004). The inclusion of a study in a meta-analysis is limited by several
factors, the most significant of which is the reporting of the information needed to compute
effect size. Very often, reports released by virtual schools and other distance education programs
do not include mean scores, comparison group scores, sample sizes, or standard deviations.
Nonetheless, the meta-analytic technique is a way to identify effects or relationships in literature
that may not be evident otherwise (Lipsey & Wilson, 2001).
Method
This quantitative synthesis is a meta-analysis of empirical studies published since 1999 that
compared the effects of web-delivered distance education with classroom-based learning on K–
12 student academic performance. Since 1999 the sophistication in the use of distance learning
tools has improved, but the types of tools available to schools have remained approximately the
same. The stages of the meta-analysis were identification and retrieval of applicable studies,
coding of study features and findings, and data analysis. These stages are described below.
For the purposes of this meta-analysis, studies were included in the analysis if they met the
following criteria for inclusion. The studies must:
• Be available as a journal article, dissertation or report in English between 1999 and 2004.
• Compare K–12 students in a distance education group to a nondistance education group,
or compare the distance education group before and after distance education.
• Use web-based telecommunications, such that at least 50 percent of the students’
participation in the course or program occurred at a physical distance from the instructor.
Meta-Analysis of Distance Education
Learning Point Associates 11
• Be quantitative, experimental, or quasi-experimental studies for which effect size could
be computed, the outcome measures were the same or comparable, and the N was 2 or
greater.
• Use student academic achievement, motivation, attitude, retention, or conduct as outcome
variables.
Location and Selection of Studies
Numerous databases, journals, websites, and bibliographic resources were searched for studies
that met the established inclusion criteria. In each case, search terms included:
 cybercharter
 cyberschool
 distance education
 distance learning
 elearning
 mlearning
 online school
 open learning
 open school
 schoolnet
 telelearning
 virtual charter
 virtual school.
Electronic searches were systematically conducted in the following databases:
 Dissertation Abstracts
 ERIC
 JSTOR
 Kluwer
 ProQuest Education
 PsychInfo
 Wilson Education.
Web searches were performed using the Google, Teoma, Grokker, MetaCrawler, and AltaVista
search sites.
Abstracts in the following distance education journals were examined:
 American Journal of Distance Education
 Computers & Education
 Distance Education
 Journal of Distance Education
 Journal of Distance Learning
 Open Learning.
Meta-Analysis of Distance Education
Learning Point Associates 12
Abstracts in the following educational technology journals were examined:
 Association for the Advancement of Computing in Education journals
 Australasian Journal of Educational Technology
 British Journal of Educational Technology
 Canadian Journal of Educational Communication
 Canadian Journal of Learning and Technology
 Computers in the School
 Educational Technology & Society
 Educational Technology Research and Development
 Journal of Computer Mediated Communication
 Journal of Computing in Childhood Education
 Journal of Educational Computing Research
 Journal of Information Technology Education
 Journal of Interactive Media in Education
 Journal of Research on Technology in Education.
Abstracts in American Educational Research Journal were examined, as were abstracts in the
following electronic journals:
 Australian Educational Computing
 Australian Journal of Educational Technology
 Electronic Journal for the Integration of Technology in Education
 International Journal of Educational Technology
 International Review of Research in Open and Distance Learning
 Journal of Asynchronous Learning Networks
 Journal of Interactive Online Learning
 Online Journal of Distance Education Administration
 TechKnowLogia: International Journal of Technologies for the Advancement of
Knowledge and Learning
 Turkish Online Journal of Distance Education.
In addition, abstracts were examined in the following conference proceedings:
 American Education Research Association
 Canadian Association for Distance Education
 EdMedia
 E-Learn/WebNet
 Society for Technology in Teacher Education.
The web sites of several distance education organizations and over 200 virtual schools were
browsed for studies, and the director of each virtual school was contacted at the email address
listed on the school’s website to request studies. The department of education website for each
state was browsed for report cards for state virtual charter schools. The reference lists of the six
recent meta-analyses of distance education shown in Table 1 were reviewed for potential studies.
Meta-Analysis of Distance Education
Learning Point Associates 13
Of the thousands of abstracts that were reviewed, 80 full-text articles, dissertations, or reports
concerning DE and traditional instruction at K–12 level were obtained and evaluated for
inclusion in the analysis. Independently, two researchers read all collected studies to determine
eligibility for inclusion based on the stated criteria. Fourteen of the studies were found to meet
all criteria for inclusion. Of the 66 studies that were examined and excluded, 28 percent were
descriptive reports, 14 percent reported on uses of telecommunications or other educational
technology that did not meet the definition of distance education, 25 percent reported results
without control or comparison group data, and 33 percent included summary data only or did not
provide data sufficient to compute effect size.
Limitations of the Review
For literature on K–12 distance education to be meaningfully synthesized, the inclusion criteria
had to be narrowly specified. This synthesis included studies with data on the performance of
grades 3–12 students in web-based distance learning programs compared to students in
classrooms. Measures of performance present in the literature do not draw a complete picture of
the full range of effects that students experience as a result of participation in distance education.
Qualitative studies, strict experimental studies, narrative reports, and other designs offer
information not acquired in this analysis. Although the inclusion criteria were designed to allow
a wide range of studies to be analyzed so that a comprehensive knowledge of K–12 distance
education would result, a small number of studies was analyzed. The results should be
interpreted with caution.
Coding of Study Features
Coding of study features allows the meta-analyst to unravel different study factors related to
variations in the phenomenon from factors related to method (Lipsey & Wilson, 2001). The
coding used in this analysis was identified from research on K–12 distance education and from
variables typically coded in contemporary meta-analyses in education. A trial conducted on a
small sample of studies led to the addition of variables in the codebook that were not present in
the initial set of variables. Each study was coded independently by two researchers according to
the established coding procedure. The full codebook is included in Appendix A. The initial interrater
agreement across all coded variables was 85 percent. Discrepancies between researchers
were discussed and resolved. The entire dataset was reviewed for the presence of discrepancies
and unexpected values.
Fourteen studies, with a total of 116 outcomes, had data sufficient to include in the analysis (see
Table 2). The dependent variable in this synthesis was student outcome measured by instruments
appropriate to the individual study given at the end of the distance education period which varied
from a few weeks to an entire academic year. The measures included district, state, or national
examinations, as well as teacher or researcher designed tests of academic performance.
The studies were coded on 45 factors, categorized into five groups: identification of studies,
distance education features, instructor/program features, study quality features, and sources of
invalidity (see Appendix A). Of particular interest were the variables associated with distance
education features (e.g. duration of the experience, role of the distance learning, role of the
Meta-Analysis of Distance Education
Learning Point Associates 14
instructor, timing of the interactions) and instructor/program features (e.g. amount of teacher
preparation for distance teaching, setting of the students). In many cases, however, the literature
failed to report the detail needed to make meaningful comparisons on these factors. The levels of
each variable were compared by computing average effect sizes for each level, but examination
of interactions among the different variables was not practical due to the small number of effect
sizes available.
Calculation of Effect Sizes
The effect sizes estimated for each study outcome were computed using Cohen’s d, defined in
this meta-analysis as the difference between the nondistance learning group and the distance
learning posttest mean scores divided by the average standard deviation. A correction factor for
small sample bias in effect size estimation (Hedges, Shymansky, & Woodworth, 1989) was used
in cases in which sample sizes were small. The unit of analysis was the study outcome. For
studies in which more than one independent group of students was evaluated, independent effect
sizes were estimated for each group, were weighted to avoid study bias, and were included in the
aggregated effect size estimate. A positive effect size, with a 95 percent confidence interval not
encompassing zero, is an indication that the distance learning group outperformed the
nondistance learning group.
Table 2
Selected study features and effect sizes for 14 studies of web-based K–12 distance education
Author, year Grade
level
Subject area School
type
Outcome
measure
Instructional
role of the
distance
learning
Timing of
interactions
N Weighted
mean
effect size
(d)
95% CI
for d
(upper/lo
wer)
Alberta
Consortium
2001*
3, 6, 9,
12
English,
mathematics,
science, social
studies
Mix of
public
and
private
National
test
Course Asynchronous 13–
397
-0.028 0.141/-
0.197
Alaska
Department of
Education and
Early
Development
2003*
4–7,
9–12
Reading,
writing,
mathematics
State
charter
State and
national
tests
Full program Synchronous 7–67 -0.005 0.303/-
0.313
Colorado
Department of
Education
2003a*
3–6 Reading,
writing,
mathematics
State
charter
State test Full program Asynchronous 33–45 -0.028 0.261/-
0.276
Colorado
Department of
Education
2003b*
7–8 Reading,
writing,
mathematics
State
charter
State test Full program Asynchronous 9–55 -0.029 0.199/-
0.258
Colorado
Department of
Education
2003c*
3–6 Reading,
writing,
mathematics
State
charter
State test Full program Combination
synchronous
asynchronous
14–23 -0.013 0.440/-
0.466
Colorado
Department of
Education
2003d*
7–8 Reading,
writing,
mathematics
State
charter
State test Full program Combination
synchronous
asynchronous
10–21 -0.013 0.449/-
0.475
Goc Karp & 9–12 Physical Public Class Portion of Asynchronous 19 -0.253 0.357/-
Meta-Analysis of Distance Education
Learning Point Associates 15
Woods 2003* education assignments course 0.863
Indiana
Department of
Education,
2004*
3, 6 Reading,
mathematics
State
charter
State test Full program Unspecified 17–18 0.001 0.470/-
0.468
Minnesota
Department of
Education
2003*
5 Reading,
mathematics
State
charter
State test Full program Unspecified 26 0.014 0.398/-
0.371
Mock 2000* 12 Science Public Teacher
made test
Portion of
course
Asynchronous 7 -0.472 0.472/-
1.416
Stevens 1999* 12 Science Public Teacher
made test
Portion of
course
Unspecified 21–33 -0.029 0.497/-
0.556
Washington
Office of the
Superintendent
of Public
Instruction
2003*
7 Reading,
Writing,
mathematics,
listening
State
charter
State test Full program Asynchronous 12–15 0.002 0.540/-
0.537
Wisconsin
Department of
Public
Instruction
2003
3 Reading State
charter
State test Full program Asynchronous 57 -0.016 0.243/-
0.276
Texas
Education
Agency 2003*
9–11 English,
mathematics,
science, social
studies
State
charter
State test Full program Combination 15–21 -0.014 0.445/-
0.474
* indicates studies yielding multiple effect sizes
Statistical Analysis of Effect Sizes
The test for heterogeneity (Q), based on Hedges and Olkin (1985), was used to determine
whether the effect sizes of the studies were homogenously distributed, in other words, to learn
whether the distribution of effect sizes around their mean was what would be expected from
sampling error alone (Lipsey & Wilson, 2001). The Q value for the weighted effect sizes was
1.485, and was considered to be homogeneous, indicating that the variance observed was likely
to be due to sampling error. Therefore, the fixed-effects model was used to estimate variance
(Kromrey & Hogarty, 2002). Study feature analyses were performed to determine the extent to
which student outcomes were moderated by the study variables. Statistical Analysis System
(SAS) software was used for the analyses. Effect size comparisons were done for the variables:
grade level, content area, duration and frequency of the distance learning experience,
instructional role of the distance education, pacing of the instruction, role of the instructor,
timing of the interactions, and types of interactions, as well as for various study quality and
invalidity factors.
Results
Characteristics of the Studies
The 14 studies included in the analysis yielded 116 independent effect sizes drawn from a
combined sample of 7561 students whose performance as a result of participation in a distance
education program was compared to control groups in which students did not participate in
Meta-Analysis of Distance Education
Learning Point Associates 16
distance education. Sixty one percent of the study results had sample sizes of less than 50, and 16
percent had sample sizes above 100. All but one of the studies included more than one
comparison, and the average number of comparisons per study was 8, ranging from one to 38.
Eighty six percent of the studies were organization reports, 7 percent were published articles, and
7 percent were dissertations. All of the studies were published between 1999 and 2004, with
eleven published during 2003 and 2004, and three published from 1999 through 2001. Two
studies were published in Canada, and the other twelve were published in the U.S.
A range of distance learning structures was examined in the literature. Half of the studies
reported on programs that used asynchronous timing in instruction. Three studies documented a
program that used a combination of synchronous and asynchronous instruction, one program was
delivered synchronously, and the remaining programs did not report on instructional timing. Ten
of the studies reported results of student participation in full year-long distance learning
programs, one included data for distance learning courses, and three studies focused on portions
of courses delivered at a distance for less than a semester. Thirteen studies included data from
programs in which students participated approximately five days per week, and the other study
did not indicate the frequency of student participation. The diversity of distance learning
structures is an indication of the wide range of educational uses to which it is being applied:
enhancement or extension to classroom instruction, school courses, and full-time educational
programs.
The studies encompassed a variety of instructional features. The bulk of the results, 75 percent,
occurred in the secondary grades, 6–12. The other results concern elementary age children, in
grades 3–5. Results from seven academic content areas were reported. Thirty percent of the
results came from tests of reading ability, followed by mathematics, which accounted for 26
percent of the results. Writing was the subject for 16 percent of the results, science was the topic
of 14 percent, and social studies made up 9 percent of the results. Three percent of results came
from physical education comparisons, and one percent from a test of listening. National tests
were used to compare outcomes in one study, state tests were used in nine studies, teacher made
tests were used in two studies, and one study reported data from both state and national tests.
Overall Effects of K–12 Distance Education
The analysis resulted in an overall weighted effect size not significantly different from zero, a
result that is consistent with the results of recent meta-analyses of distance education (see Table
1), which tend to show that distance education is as effective as classroom instruction. The
weighted mean effect size across all results was -0.028, with a standard error of 0.045 and a 95
percent confidence interval from -0.116 to 0.060. The average unweighted Cohen’s d was -0.034,
and the median effect size was -0.015. The effect sizes varied considerably among the studies.
Figure 1 displays the full range of effect sizes calculated for the 116 results across the horizontal
axis, and the number of results having each effect size on the vertical axis. The spike in the
number of results around the zero effect size is an indicator of the tendency of the overall effect
size. Unweighted effect sizes ranged from -1.158 to 0.597, with a standard deviation of 0.157,
indicating that some applications of distance education appeared to be much better than
classroom instruction and others were much worse.
Meta-Analysis of Distance Education
Learning Point Associates 17
Distribution of unweighted effect sizes
0
20
40
60
80
100
-1.4 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8
Effect size
Number of results
Figure 1. Distribution of unweighted effect sizes of 116 outcomes
The 95 percent confidence intervals also show wide variability in their size, as displayed in
Figure 2. Only one confidence interval did not encompass zero, and all but three effect sizes fell
between 0.5 and -0.5. Each of the fourteen studies and all except one of the 116 outcomes within
the studies had individual effect sizes that did not differ significantly from zero, indicating that in
almost every comparison, students in distance education programs performed as well as students
in classroom-based programs.
Meta-Analysis of Distance Education
Learning Point Associates 18
Figure 2. 95 percent confidence intervals for individual effect sizes of 116 outcomes.
Of the 45 factors coded in the study, the following 30 were examined to determine sources of
significant variation in effect sizes. Ten of the remaining variables were used for identifying the
studies or computing effect size, and the other five could not be compared because the studies
did not include the data for coding the variables, or the variable was not a relevant factor in the
studies. The variables that went uncoded due to the absence of data were the frequency of
student participation in distance learning, the level of preparation of the teachers in distance
education, and the amount of experience of the teachers in distance education. The variables that
were not relevant factors for the studies were control for the effects of a second testing, and
control for the effects of a pretest. Analysis of variance was not meaningful for some of the
variables because of missing data in the studies, resulting in a high number of cases in which a
value of “unspecified” was coded for the variable.
Publication and Methodological Variables
Twenty variables were coded to discover whether publication or methodological variables
accounted for variation in effect sizes. The publication features included the year of publication,
the type of publication, and the region of publication. The methodological variables related to the
testing sequence in the study, the type of achievement measure used in achievement studies,
pretest equivalency measures, study design, statistical power, and control for 12 potential sources
95 PercentConfidence Intervals for Individual Effect Sizes
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
Meta-Analysis of Distance Education
Learning Point Associates 19
of invalidity. None of the variable comparisons resulted in effect sizes significantly different
from zero (see Table 2 and Figure 2).
Distance Education Variables
Eleven variables were used to identify the features of the distance education experience that may
play a role in student performance. They were duration of the program, frequency of use of
distance learning, instructional role of the program, number of distance learning sessions,
duration of distance learning sessions, pacing of the instruction, role of the instructor, timing of
the interactions, type of interactions, amount of teacher preparation for distance instruction, and
amount of teacher experience in distance instruction. Because of the individualized nature of
distance education, only two of the studies indicated specific numbers and durations of distance
learning sessions, and they were studies of limited partial-course experiences. Half of the studies
did not indicate whether students or instructors set the pace within the distance learning
timeframe, while three of the programs were completely self-paced, and four were designed for
students to set their pace within parameters set by the instructor. In terms of the role of the
instructor in teaching, one program was fully moderated, five were nonmoderated, four used a
combination of moderated and nonmoderated activities, and four did not indicate the instructors’
role. Ten programs used a combination of interactions among students, content, instructors, and
others; one limited interactions to student-content; and three did not specify interaction types. No
studies described the levels of instructor preparation or experience required of or possessed by
the instructors. All levels of each distance education variable had effect sizes not significantly
different from zero.
Instructional and Program Variables
The five variables that indicated the extent to which instructional and program factors played a
role in student outcomes were grade level, school type, content area, the qualifications of the
teacher in the teaching field, and the setting of the students. Twelve of the studies indicated that
the instructors were certified teachers, and the other two studies did not describe the credentials
of the instructors. In five of the programs, students participated from home or a nonschool
location, four programs are designed such that students completed some work from home and
some in a school setting, in three programs, students did their distance learning work while at a
school, and two programs did not specify the setting of the students. All instructional and
program factors had effect sizes that were effectively zero.
Discussion
The literature reviewed in this meta-analysis includes results from 116 comparisons of grades 3–
12 web-based distance education programs with classroom-based teaching, including data for
7561 students. The questions of the effectiveness of distance education for K–12 student
performance, and of the factors influencing its effectiveness were addressed using fixed-effects
effect size estimation. The findings confirm those of other recent meta-analyses of distance
education programs, and provide a needed update to the meta-analysis focused on K–12 students
which was completed in 1998 just as the web-based systems were beginning to be studied in
virtual schooling. The analysis showed that for the factors examined, distance learning did not
Meta-Analysis of Distance Education
Learning Point Associates 20
outperform or underperform classroom instruction. The number of studies was small, and many
studies did not report detailed information, so the results should be viewed as indications of
tendencies rather than prescriptions for practice. What has been learned from these results is that,
based on the best research available on online K–12 distance education programs, such programs
are effective for student learning. Prior to this point, the field has relied on small individual
studies, syntheses that included outdated analog technology, and syntheses that included adult
learners.
Implications for Research and Practice
Distance education as it has been implemented at the K–12 level over the past decade has
improved over time according to several measures: providing access to education and choice in
course offerings to increased numbers of students, offering education to a larger range of grade
levels and ability levels, using more interactive and widely accessible technologies, and leading
students to academic success on a wider range of achievement instruments. The effect of
distance education on learning may be moderated by several factors, existing as it does in a very
complex web of educational, technological, and social dynamics. Factors such as the design of
the distance learning system, the demands of the content, the abilities and disabilities of the
student, and the quality of the teacher are likely to be influential factors, as they are in
conventional educational enterprises. The consistency of the effects shown in the studies
analyzed in this review suggest that as distance education is currently practiced, educators and
other stakeholders can reasonably expect learning in a well-designed distance education
environment to be equivalent to learning in a well-designed classroom environment.
How will K–12 distance education realize greater potential and maximize it effectiveness? How
will designers and managers of K–12 distance education programs make better decisions in order
to design and deliver a more effective program? The answers lie in changes in the ways
policymakers and researchers do their work in this complex context. In order for distance
education to be evaluated, data must be collected and reported in detail. Such data collection
begins with identification of goals. Policymakers and evaluators must enter into a partnership in
which common goals are identified, an evaluation plan is acted on, and detailed reporting
follows. Evaluation must be seen as a tool to support policy setting and decision making (Means
& Haertel, 2004). It is no longer enough to ask whether distance education is effective, we need
to understand why (Sabelli, 2004). We need to know how to make it more effective, what factors
contribute most to effectiveness, and in what contexts the factors operate. Acquiring this
knowledge requires consensus on a definition of effectiveness that goes beyond standardized
tests, and a system for identifying and measuring factors that influence effectiveness. As Means
and Haertel stress, “many studies of the effects of technology-supported innovations are hindered
by a lack of measures of student learning commensurate with the initiative’s goals” (p. 99).
One factor warranting special consideration in assessing the effectiveness of virtual schooling is
teacher quality. In classrooms, teacher effectiveness is a strong determiner of differences in
student learning, far outweighing differences in class size and heterogeneity (Darling-Hammond,
2000). Based on the similarities in student outcomes between distance and classroom learning,
there is every reason to expect that teacher preparation is critical in distance education. However,
there has been very little formal preparation available addressing the unique nature of online
Meta-Analysis of Distance Education
Learning Point Associates 21
instruction and very little time for teachers to develop their expertise as online instructors. As
professional development becomes more common and expertise grows, student success is likely
to grow as well.
As second factor that is growing in importance in K–12 distance education is the emergence of
virtual charter schools. By 2002, there were about 2000 charter schools nationwide, and the No
Child Left Behind Act allows public schools that “chronically fail” to make adequate yearly
progress to be restructured as charter schools (Nelson, Rosenburg, & Van Meter, 2004, p. 1).
According to state department of education websites, there are now almost 100 virtual charter
schools operating. This synthesis includes data from ten virtual charter schools, all of which
performed at levels equivalent to nonvirtual public schools in their states. In contrast, the 20004
report on charter school achievement on the National Assessment of Education Progress (Nelson
et al) provides evidence that charter schools overall are underperforming when compared to
noncharter public schools. Charter school students had significantly lower achievement in grades
4 and 8 math and reading, even when eligibility for free or reduced price lunch and urban
location were factored into the comparison. When minority status was used as a factor, it was
found that black and Hispanic charter school students scored lower in 4th grade math and
reading, but the difference was not significant. The fact that virtual charter school students were
not shown to score lower than nonvirtual school students in this meta-analysis is an indicator of
the success of distance education for K–12 learning.
Teacher quality and classification as a charter school have been recognized as factors that can
influence student learning in classrooms, but little data is available about the influences of these
factors in virtual schooling. Practitioners and policymakers in K–12 distance education are urged
to use data-driven decision making, and to do so they must be informed by experience and data
must be available. In 2004, there have been fewer than ten years of accumulated experience and
too little detailed research published on web-based distance education methods. The lack of
detail in the research to date hinders thorough investigation of the factors influencing practice,
and limits what can be learned for the improvement of practice. A coordinated research and
reporting effort is needed in order to improve the cycle of conducting research on practice and
applying research to improve practice.
Conclusions
Students can experience similar levels of academic success while learning using
telecommunications and learning in classroom settings. While distance learning as it is practiced
in today’s virtual schools uses technology that is less than ten years old and advances rapidly, the
literature has shown that a student’s education online can be as effective as it is in a classroom,
provided that a classroom with the appropriate course is accessible to the student. As the power
of communication technology and educational technology grow, the skill of distance educators
and designers will be challenged to provide experiences that use that power to provide an
experience for students that improves on classroom instruction with its limits of time and place.
Research in K–12 distance education is maturing alongside the technology and those who use it,
but current web-based distance education systems have only been studied for about the last five
years at the K–12 level, a very short time in which to build a body of literature.
Meta-Analysis of Distance Education
Learning Point Associates 22
This meta-analysis represents an investigation into the literature on K–12 web-based distance
education with attention on the factors likely to influence student performance. The result shows
variation in the degree of success students have experienced, and a need for more information if
firm conclusions are to be drawn. Blomeyer (2002) stated the recommendation well: “Support
for additional professionally designed and executed program evaluations and scientific
educational research should be given a high priority in all public and private agencies supporting
effective implementation and use of online learning in K–12 learning communities” (page 10).
The importance of knowledge about effective virtual schooling cannot be overstated, because of
the current boom in the numbers of virtual schools and students, and because of the essential role
virtual schools can play in school reform movements and workforce development efforts. As of
spring, 2004, there were roughly 2,400 publicly-funded cyber-based charter schools and state
and district virtual schools in 37 U.S. states, with an estimated 40,000 to 50,000 students
participating in online courses, according to Susan Patrick, Director of the U.S. Department of
Education's Office of Educational Technology (Fording, 2004). With recent and continued
growth in virtual schools, virtual school leaders and policy makers will need a strong research
foundation on which to base decisions.
Several groups in the U.S. have identified school reform, particularly high school reform, as
priorities in coming years. The U.S. Department of Education has identified high school reform
models that support student achievement, and has recognized small school size, scheduling
choice, charter schools, career academies, early college initiatives, and student engagement as
research-based models that contribute to improved student achievement (U.S. Department of
Education, 2004). The National Governor’s Association has formed a task force to study
redesigning high schools in order to make them “more rigorous and relevant to the lives of
America’s youth” (National Governor’s Association, 2004). The task force initiative responds to
employers’ needs for more skilled and better educated workers by suggesting that reforms
include choices in high school programs and opportunities to earn college credit or professional
credentials. The National Association of Secondary School Principals in 2004 published
Breaking Ranks II, which calls for reforming high schools to become more rigorous and
personalized (National Association of Secondary School Principals, 2004), and the National
High School Alliance has developed the Catalog of Research on Secondary School Reform. The
catalog compiles studies of effective school reform programs, including those based on early
college, smaller schools, student interests and learning styles, at-risk student needs, talent
development, and career academies (National High School Alliance, 2004). Each of the reform
models described and recommended by these groups is an example of a strength that has been
shown by virtual schools. By offering scheduling flexibility, personalization, freedom from a
large physical school, engaging tools of distance learning, opportunities to accelerate learning,
and access to rigorous academic programs, virtual schools are not just important examples of
school reform models, but virtual schools may represent the best hope for bringing high school
reform quickly to large numbers of students.
Another strength of virtual schools is their unique capability for immersing students in
information and communication technologies (ICT). An international effort is underway to
improve ICT literacy as a “contribution to the development of human capital” (Educational
Testing Service, 2001, p. iii). An international panel convened by the Educational Testing
Meta-Analysis of Distance Education
Learning Point Associates 23
Service determined that ICT skill is needed by citizens to function in the current technological
climate, and that ICT skills are needed to help people worldwide meet fundamental needs,
making ICT literacy a global objective. The development of ICT literacy begins with access to
technology, and many publicly-funded virtual schools have found ways to bridge the access
divide by providing computers to students. Virtual school students must develop ICT skills to be
successful in online learning, and they may become the sought-after employees of the near
future. Because of the global need for ICT skills and their role in virtual schools, demand could
rise for data on effective virtual schools as more are developed worldwide.
The Need for Prospective Study in Virtual Schooling
An important step toward improving the state on virtual schools research was taken in 2004
when the U.S. Department of Education hosted an E-learning Summit to explore the status of K–
12 e-learning in the U.S. The DOE Office of Educational Technology is showing leadership by
identifying e-learning as a priority in the new National Educational Technology Plan.
Technology, including e-learning, is seen as a force that can transform education because of the
power of e-learning to individualize, personalize and differentiate instruction. Plans for the
Federal role in e-learning leadership will include development of an e-learning clearinghouse
listing programs for students, a process for addressing quality and accreditation issues, and
support for developing online content. Such initiatives begin to bring knowledge and expertise to
more stakeholders, assist policymakers and practitioners in accessing information, and serve as a
focal point for guiding future work that will improve outcomes across the spectrum.
As a relatively recent innovation in the sometimes slow-moving world of education, distance
education has been shown over decades with every variety of technology to work effectively
although it works in very different ways than classroom instruction does, it meets different
needs, and serves different audiences, having had far less time in which to mature, as evidenced
by the studies included in this meta-analysis. The literature contains reports on distance
education programs in which student outcomes exceed those in conventional classrooms (see
citations in “Distance Education in the K–12 Context” section), but in order to make use of such
data in syntheses such as this one, complete data need to be reported.
Recommendations for K–12 Online Learning Policy and Practice
Policy-makers and practitioners should continue to move forward in developing and
implementing K–12 distance education programs when those programs meet identified needs
and when they are designed and managed as carefully as traditional education programs. The “no
significant difference” result reported here and elsewhere lends confidence to distance educators
that their ongoing efforts are likely to be as effective as classroom-based education. This
synthesis, considered together with current policy and recent research findings, demonstrates that
students of many types and ages can learn in many content areas using the flexibility and choices
afforded by distance education. In their recent article, New Millenium Research for Educational
Technology: A Call for a National Research Agenda, Roblyer and Knezek (2003) recommended
a focus and priorities for a future technology research agenda. The focus, they stated, should be
providing a rationale for technology use. The priority is to explain why students and educators
should use technology.
Meta-Analysis of Distance Education
Learning Point Associates 24
Optimally, the research on K–12 distance education would recommend specific practices that
would lead to results that exceed those in conventional education settings. The barriers that
prevent such recommendations include:
 a limit on the educational expertise focused on distance education as an area of study. A
small subset of educational researchers have elected to focus on virtual schooling, either
as doctoral candidates, faculty, program directors or independent evaluators.
 a rather short-sighted view of the purposes of distance education, a lack of consensus
about the goals of distance education, and an accompanying lack of evaluation directed at
assessing progress toward those goals. Distance education has been seen primarily as a
substitute for classroom instruction, rather than a potentially more effective way of
learning. Until the goal is established of reaching a higher potential, research will
continue to determine whether distance education is as effective as classroom instruction,
rather than looking for ways that distance education can excel.
 a failure to take into account the complexity of systems in which distance education
operates. Complexity is difficult to quantify, but virtual schooling evaluation and
research can begin to track a greater range of influences, leading to a more thorough
understanding of its effects.
 a paucity of research and reporting that includes details sufficient for quantitative
synthesis. Most reports on virtual schooling released in the past omit sample sizes, mean
scores, standard deviations, and other details needed for big-picture synthesis.
For distance education to add a prospective agenda to the archive of valuable retrospective study
that currently guides the field, five major action recommendation must be addressed by online
learning practitioners, online learning district-level leadership, and Federal and State educational
policy makers:
1. First, the broader educational community needs to become better informed about K–12
online learning and distance education, to foster better communication among the widest
range of experts and practitioners who have the potential to contribute to advances in the
field.
This crucial informational campaign requires professionals working in distance education
in any capacity to network by participating in conferences, publishing articles and papers,
and contributing to discussions locally and globally where people who are not involved in
distance education can learn.
2. Second, the community of distance education policy makers, researchers, and
practitioners should develop and articulate a long-range view of the intended and
expected benefits of distance education and become advocates for suitably long-term
studies of its effects.
The list of potential benefits should be broad, and should be a close match to the benefits
or “effects” anticipated for any educational experience. Curriculum content should
include a liberal education in which knowledge, skills, and dispositions are developed
that successful students need in order to enjoy a full life in a democracy. But effects and
Meta-Analysis of Distance Education
Learning Point Associates 25
benefits should also include academic literacies, technology skills, and academic
standards.
This list of crucial, performance-based knowledge, skills, and dispositions must serve as
a guide in the stages of design, implementation, and evaluation of programs. Consensus is
needed on the goals of distance education, and plans should follow to evaluate progress
toward those goals. Distance education program directors should see researchers as
partners in informing practice.
3. Third, because education occurs in a dynamic context, and the rapid change in the
technology used in distance education adds to the complexity, evaluation of distance
education programs needs to account for more of this complexity than has so far been the
practice.
A common “codebook” or heuristic descriptive system should be created and refined to
ensure that outcomes from distance and online learning programs can be accurately
compared to other online and distance programs and to face-to-face instruction. A
descriptive system supporting comparative analysis of all varieties of traditional and
online and distance learning delivery systems will dramatically increase both the
generalizability of results and the synthesizability of research findings available to inform
development, implementation and institutionalization of online and distance learning
programs.
4. Finally, standards are needed for reporting the academic and programmatic outcomes of
distance education programs. Many K–12 distance education program directors collect
admirable amounts of data, and conduct in-house analyses, but until there are standards
set to guide the reporting of data, educational research will remain limited to examining
results from only a small, enlightened subset of these programs.
5. The actions recommended require coordination and leadership. Leadership should begin
at the national level and include professional organizations like the North American
Council on Online Learning and International Society for Technology in Education. The
United States Department of Education and the leading professional organizations and
groups should assume a leadership role organizing a national distance learning and online
learning community of practice to work toward enacting these essential action
recommendations.
Distance educators belong to a wide variety of overlapping professional groups and
associations that have the potential to contribute to a powerful and effective coalition.
The larger coalition needed to weld a broader professional consensus should serve as a
central clearinghouse for information about K–12 online and distance education, a
matchmaking service for programs and evaluators, and as an organizational focus for
organizing national efforts to support online and distance learning policy, program
development, and professional development.
Meta-Analysis of Distance Education
Learning Point Associates 26
Learning, progress, and data-driven decisions require the availability of relevant data. The K–12
distance education and online learning communities certainly have the infrastructure for sharing
that information. What is needed now is an adequate and uniform system for describing
academic and programmatic outcomes within and across a variety of programs and instructional
delivery systems, and uniform metrics and standards that can support comparisons within and
across the various delivery systems and instructional modalities.
With ubiquitous availability of good information on the performance of all K–12 educational
programs and instructional systems, parents and practitioners, policymakers and national
political leadership will be able to make the very best informed decisions about how to best
educate and equip all our children for life and success during the ensuing twenty-first century.
Meta-Analysis of Distance Education
Learning Point Associates 27
References
References marked with an asterisk indicate studies used in the meta-analysis.
Allen, M., Bourhis, J., Burrell, N., & Mabry, E. (2002). Comparing student satisfaction with
distance education to traditional classrooms in higher education: A meta-analysis. The
American Journal of Distance Education, 16, 83–97.
*Alaska Department of Education and Early Development. (2003). Spring 2003 benchmarks,
Delta/Greely school district - Delta Charter Cyber School. Retrieved July 19, 2004, from
http://www.eed.state.ak.us/tls/assessment/AsmtVer/SchoolAsmtVerSuptSearch.cfm
Arkansas Virtual School. (2003). 2002–2003 testing overview. Dardanelle, AR: author.
Bangert-Drowns, R., Hurley, M., & Wilkinson, B. (2004). The effects of school-based writingto-
learn interventions on academic achievement: A meta-analysis. Review of Educational
Research, 74 (1), 29–58.
Barker, K., & Wendel, T. (2001). E-learning: studying Canada's virtual secondary schools.
Kelowna, BC: Society for the Advancement of Excellence in Education. Retrieved April
11, 2004, from http://www.excellenceineducation.ca/pdfs/006.pdf
Bernard, R. M., Lou, Y., Abrami, P.C., Wozney, L., Borokhovski, E., Wallet, P.A., Wade, A., &
Fiset, M. (2003, April). How does distance education compare to classroom instruction? A
meta-analysis of the empirical literature. Symposium conducted at the annual meeting of
the American Educational Research Association, Chicago, IL.
Bianchi, W. (2002). The Wisconsin school of the air: Success story with implications.
Educational Technology & Society 5 (1).
Blomeyer, R. (2002). Virtual schools and e-learning in K–12 environments: Emerging policy and
practice. NCREL Policy Issues, 11. Naperville, IL: North Central Regional Educational
Laboratory. Retrieved May 2, 2004, from
http://www.ncrel.org/policy/pubs/pdfs/pivol11.pdf
Bolton, J. (2002). Web-based distance education: Pedagogy, epistemology, and instructional
design. University of Saskatchewan. Retrieved April 21, 2004, from
http://www.usask.ca/education/coursework/802papers/boulton/boulton.pdf
Bond, A. (2002). Learning music online: An accessible program for isolated students.
Kensington Park, SA: Australian National Training Authority. Retrieved April 12, 2004,
from http://www.ncver.edu.au/research/proj/nr1013.pdf
Bransford, J., Brown, A., & Cocking, R. (Eds.). (1999). How people learn. Washington, DC:
National Academies Press.
Meta-Analysis of Distance Education
Learning Point Associates 28
Brooks, J. & Brooks, M. (1993). The case for constructivist classrooms. Alexandria, VA:
Association for Supervision and Curriculum Development.
Bruckman, A. (1998). Computer supported cooperative work. The Journal of Collaborative
Computing, 7, 47–86.
Calderoni, J. (1998). Telesecundaria: Using TV to bring education to rural Mexico. Education
and Technology Technical Notes Series: World Bank Human Development Network.
Retrieved April 8, 2004, from
http://wbln0018.worldbank.org/HDNet/HDdocs.nsf/C11FBFF6C1B77F9985256686006D
C949/1635F1703FE053B385256754006D8C3F/$FILE/telesecundaria.pdf
Cavanaugh, C. S. (2001). The effectiveness of interactive distance education technologies in K–
12 learning: A meta-analysis. International Journal of Educational Telecommunications,
7(1), 73–88.
* Colorado Department of Education. (2003a). Branson alternative school, grades 1–6 school
accountability report. Denver, CO: Author. Retrieved July 19, 2004, from
http://reportcard.cde.state.co.us/reportcard/pdf/2003_1750_0948_E.pdf
* Colorado Department of Education. (2003b). Branson alternative school, grades 7–8 school
accountability report. Denver, CO: Author. Retrieved July 19, 2004, from
http://reportcard.cde.state.co.us/reportcard/pdf/2003_1750_0948_M.pdf
* Colorado Department of Education. (2003c). Connections academy, grades 1–6 school
accountability report. Denver, CO: Author. Retrieved July 19, 2004, from
http://reportcard.cde.state.co.us/reportcard/pdf/2003_0880_1887_E.pdf
* Colorado Department of Education. (2003c). Connections academy, grades 7–8 school
accountability report. Denver, CO: Author. Retrieved July 19, 2004, from
http://reportcard.cde.state.co.us/reportcard/pdf/2003_0880_1887_M.pdf
Conzemius, A; & Sandrock, P. (2003). Developing world language programs in elementary
grades. Madison, WI: Wisconsin State Department of Public Instruction. (ERIC Document
Reproduction Service No. ED 480156)
Darling-Hammond, L. (2000). Teacher quality and student achievement: a review of state policy
evidence. Education Policy Analysis Archives, (8)1.
Dees, S. (1994). An investigation of distance education versus traditional course delivery using
comparisons of student achievement scores in advanced placement chemistry and
perceptions of teachers and students about their delivery system (satellite course).
Unpublished dissertation, Northern Illinois University, DeKalb.
Meta-Analysis of Distance Education
Learning Point Associates 29
Duffy, T., & Cunningham, D. (1996). Constructivism: Implications for the design and delivery
of instruction. In Jonassen, D. (Ed.), Handbook of research on educational
telecommunications and technology. New York: Macmillan Library Reference.
Duffy, T., & Jonassen, D. (Eds.). (1992). Constructivism and the technology of instruction: A
conversation. Hillsdale, NJ: Lawrence Erlbaum Associates, Publisher.
Educational Testing Service. (2001). Digital transformation: A framework for ICT literacy.
Retrieved October 9, 2004, from http://www.ets.org/research/ictliteracy/ictreport.pdf
Falck, A-K; Husu, J; Kronlund, T., Kynaslahti, H., Salminen, J., & Salonen, M. (1997). Testing
virtual classroom in the school context. Distance Education. 18, (2), 213.
Fording, L. (2004). Education, 21st century-style. Newsweek. Retrieved September 9, 2004, from
http://www.msnbc.msn.com/id/4633126/
Gershaw, D. (1989). Line on life: locus of control. Retrieved May 29, 2004, from
http://www3.azwestern.edu/psy/dgershaw/lol/ControlLocus.html
* Goc Karp, G., & Woods, M. (2003). Wellness NutriFit online learning in physical education
for high school students. Journal of Interactive Online Learning, 2 (2). Retrieved April 9,
2004, from http://www.ncolr.org/jiol/archives/2003/fall/03/index.html
Hedges, L., Shymansky, J., & Woodworth, G. (1989). Modern methods of meta-analysis.
Washington, DC: National Science Teachers Association.
Hinnant, E. (1994). Distance learning using digital fiber optics: a study of student achievement
and student perception of delivery system quality. Unpublished dissertation, Mississippi
State University, Starkeville.
* Indiana Department of Education. (2004). School snapshot: Irvington community school.
Indianapolis, IN: Author. Retrieved July 20, 2004, from
http://mustang.doe.state.in.us/SEARCH/snapshot.cfm?schl=1537
Jordan, A. (2002). An investigation into the effects of online teaching and learning on
achievement outcomes at the secondary level. Unpublished dissertation, Fayetteville State
University, Fayetteville, NC.
Kearsley, G. (2000). Online education. Belmont, CA: Wadsworth.
Keegan, D. (1996). Foundations of distance education. London: Routledge.
Kozma, R., Zucker, A., Espinoza, C., McGhee, R., Yarnall, L., Zalles, D., & Lewis, A. (2000).
The online course experience: evaluation of the virtual high school’s third year of
implementation, 1999-2000. Arlington, VA: SRI. Retrieved December 8, 2003, from
http://www.govhs.org/Images/SRIEvals/$file/SRIAnnualReport2000.pdf
Meta-Analysis of Distance Education
Learning Point Associates 30
Kromrey, J., & Hogarty, K. (2002). Estimates of variance components in random effects metaanalysis:
sensitivity to violations of normality and variance homogeneity. American
Statistical Association Joint Statistical Meeting, 1963–1968.
Lipsey, B., & Wilson, D. (2001). Practical meta-analysis. Thousand Oaks, CA: SAGE
Publications, Inc.
Machtmes, K., & Asher, J. W. (2000). A meta-analysis of the effectiveness of telecourses in
distance education. The American Journal of Distance Education, 14 (1), 27–46.
McGreal, R. (1994). Comparison of the attitudes of learners taking audiographic
teleconferencing courses in secondary schools in northern Ontario. Interpersonal
Computing and Technology, 2 (4), 11–23. Retrieved April 12, 2004, from
http://www.helsinki.fi/science/optek/1994/n4/mcgreal.txt
Means, B., & Haertel, G. (2004). A blueprint for a national research agenda to evaluate
educational technology. In Means, B. & Haertel, G. (Eds.). Using Technology Evaluation
to Enhance Student Learning. New York: Teachers College Press.
Mills, S. (2002). School isn’t a place anymore: an evaluation of virtual Greenbush online
courses for high school students. University of Kansas: Schiefelbusch Institute for Life
Span Studies. Retrieved May 2, 2004, from
http://media.lsi.ku.edu/research/vgeval/vgevaluationreport.pdf
* Minnesota Department of Education. (2003). 2003 report card: cyber village academy.
Minneapolis, MN: Author. Retrieved July 21 2004, from
http://education.state.mn.us/ReportCard/2003/RCF402507010.pdf
* Mock, R. (2000). Comparison of online coursework to traditional instruction. Unpublished
thesis, Michigan State University, East Lansing. Retrieved April 29, 2004, from
http://hobbes.lite.msu.edu/~robmock/masters/abstract.htm
National Association of Secondary School Principals. (2004). Breaking ranks II: Strategies for
leading high school reform. Retrieved October 9, 2004, from
http://www.nassp.org/breakingranks/breakingransk2.cfm
National Association of State Boards of Education. (2001). Any time, any place, any path, any
pace: Taking the lead on e-learning policy. Alexandria, VA: author.
National Governor’s Association. (2004). Governors leading high school reform efforts.
Retrieved October 9, 2004, from
http://www.nga.org/nga/legislativeUpdate/1,1169,C_ISSUE_BRIEF^D_7342,00.html
National High School Alliance. (2004). Catalog of research on secondary school reform.
Retrieved October 9, 2004, from http://www.hsalliance.org/research/index.asp
Meta-Analysis of Distance Education
Learning Point Associates 31
Nelson, F., Rosenburg, B. & Van Meter, N. (2004). Charter school achievement on the 2003
national assessment of educational progress. Washington, DC: American Federation of
Teachers. Retrieved September 9, 2004, from
http://www.aft.org/pubs-reports/downloads/teachers/NAEPCharterSchoolReport.pdf
Roblyer, M. & Knezek, G. (2003). New millennium research for educational technology: A call
for a national research agenda. Journal of Research on Technology in Education, 36 (1),
60–71.
Rotter, J. B. (1989). Internal versus external control of reinforcement: A case history of a
variable. American Psychologist, 45, 489–493.
Ryan, W. (1996). The effectiveness of traditional vs. audiographics delivery in senior high
advanced mathematics course. Journal of Distance Education/Revue de l'enseignement à
distance (11)2. Retrieved April 13, 2004, from http://www.cade-aced.ca/en_pub.php
Sabelli, N. (2004). Policy, planning, and the evaluation of learning technology. In Means, B. &
Haertel, G. (Eds.). Using Technology Evaluation to Enhance Student Learning. New York:
Teachers College Press.
* Schollie, B. (2001). Student achievement and performance levels in online education research
study. Edmonton, Alberta: Alberta Online Consortium. Retrieved April 25, 2004, from
http://www.albertaonline.ab.ca/pdfs/AOCresearch_full_report.pdf
Shachar, M. & Neumann, Y. (2003). Differences between traditional and distance education
academic performances: a meta-analytic approach. International Review of Research in
Open and Distance Learning. October.
Slavin, R. (2003). Educational psychology theory and practice, 7th Ed. Boston, MA: Pearson
Education, Inc.
Southern Regional Education Board. (2003). Essential principles of high-quality online teaching:
guidelines for evaluating K–12 online teachers. Atlanta, GA: author. Retrieved May 2,
2004, from http://www.sreb.org/programs/EdTech/pubs/PDF/Essential_Principles.pdf
* Stevens, K. (1999). Two Canadian approaches to teaching biology, chemistry, mathematics
and physics to senior high school students in virtual classes. Australasian Science
Education Research Association. Retrieved April 9, 2004, from
http://www.tellearn.mun.ca/pubs/virtual.html
* Texas Education Agency. (2003). 2002–2003 campus performance: Southwest Virtual
Preparatory School. Austin, TX: Author. Retrieved July 22, 2004, from
http://www.tea.state.tx.us/perfreport/aeis/2003/campus.srch.html
Meta-Analysis of Distance Education
Learning Point Associates 32
Ungerleider, C.S., & Burns, T.C. (2003). A systematic review of the effectiveness and efficiency
of networked ICT in education: A state of the field report to the Council of Ministers of
Education, Canada and Industry Canada. Unpublished report.
U.S. Department of Education. (2004). High school reform models that support student
achievement. Retrieved October 9, 2004, from
http://www.ed.gov/about/offices/list/ovae/pi/hs/reform.html
* Washington Office of the Superintendent of Public Instruction. (2003). Internet academy 2002–
2003 WASL results. Olympia, WA: Author. Retrieved July 22, 2004, from
http://reportcard.ospi.k12.wa.us/Reports/summary.aspx?schoolId=1164&reportLevel=School
Wedemeyer, C.A. (1981). Learning at the back door: Reflections on non traditional learning in
the lifespan. Madison: University of Wisconsin Press.
* Wisconsin Department of Public Instruction. (2003). Successful school guide: Wisconsin
Virtual Academy. Madison, WI: Author. Retrieved July 22, 2004, from
http://data.dpi.state.wi.us/data/questions.asp?fullkey=011945040100&DN=Northern+Ozau
kee&SN=Wisconsin+Virtual+Academy&TYPECODE=6&CTY=45&ORGLEVEL=SC
Yasin, K. & Luberisse, Y. (1997). Meeting the needs of a new democracy: Multichannel learning
and interactive radio instruction in Haiti: a case study. Washington, DC: USAID.
Retrieved April 29, 2004, from http://ies.edc.org/pubs/book11.htm
Meta-Analysis of Distance Education
Learning Point Associates 33
Appendix
Coded Variables and Study Features in the Codebook
A. Identification of studies
1. Study number (“study”).
2. Finding/hypothesis number (“finding”).
3. Author name (“author”). Last name of first author.
4. Year of publication (“year”).
5. Number of findings/hypotheses within study (“number”).
6. Country (“country”).
Unspecified=0,
USA=1,
Canada=2,
Mexico/Central America/South America=3,
Europe=4,
Asia=5,
Africa=6,
Australia/Pacific=7,
Multinational=8,
Other=9.
7. Grade level of students (“grade”).
Unspecified=00,
grades 1–12 use 01 to 12,
Mixed primary (K–2) =13,
Mixed intermediate (3–5) =14,
Mixed middle (6–8) =15,
Mixed high (9–12) =16,
K–12=17,
other=18.
8. School type (“school”).
Unspecified=0,
Public district sponsored=1,
Public state sponsored=2,
Private=3,
Other=4,
Charter=5,
Combination=6.
Meta-Analysis of Distance Education
Learning Point Associates 34
9. Content area (“content”).
Unspecified=0,
Reading/language arts=1,
Mathematics=2,
Social studies=3,
Science=4,
Computers/technology=5,
Foreign language=6,
Arts=7,
Physical education=8,
Other=9,
Writing=10.
10. Type of publication (“publication”).
Published journal article=1,
Journal article in press=2,
Book chapter=3,
Report=4,
Dissertation=5,
Conference paper=6.
B. Distance learning features
1. Duration of distance learning experience (“duration”).
Less than one semester=1,
One semester=2,
More than one semester=3.
2. Frequency of distance learning experience (“frequency”).
Unspecified=0,
From 5 to 7 days per week=1,
From 1 to 4 days per week=2,
From 1 to 3 days per month=3,
Less than monthly=4.
3. Instructional role of distance learning (“role”).
Unspecified=0,
Full-time educational program=1,
Courses to supplement an educational program or partial educational program=2,
Supplement to a specific course=3.
4. Number of distance learning sessions (“dlnumber”).
Unspecified=0,
List number of sessions.
Meta-Analysis of Distance Education
Learning Point Associates 35
5. Duration of distance learning sessions (“dlduration”).
Unspecified=0,
List average minutes per session.
6. Pacing of distance learning instruction (“pacing”).
Unspecified=0,
Completely self-paced=1,
Student sets pace within instructor-determined parameters=2,
Pacing completely specified by program or instructor=3.
7. Instructor role (“instructrole”).
Unspecified=0,
Fully moderated=1,
Nonmoderated=2,
Combination=3,
Other=4.
8. Timing of interactions (“timing”).
Unspecified=0,
Synchronous=1,
Asynchronous=2,
Combination=3,
Other =4.
8. Type of interactions (“interaction”).
Unspecified=0,
Student—content=1,
Student—instructor=2,
Student—student=3,
Student—others=4,
Combination=5,
Other=6.
C. Instructor/program features
1. Amount of teacher preparation in distance learning (“instructprep”).
Unspecified=0,
List hours of preparation.
2. Amount of teacher experience in distance learning (“instructexp”).
Unspecified=0,
List years of experience.
Meta-Analysis of Distance Education
Learning Point Associates 36
3. Qualifications of teacher in the teaching field (“instructqual”).
Unspecified=0,
Certified in content area=1,
Certified but teaching out of field=2,
Alternative or provisional certification=3,
Uncertified=4,
Other=5.
4. Setting of students during distance learning (“setting”).
Unspecified=0,
Home=1,
School=2,
Other=3,
Combination=4.
D. Study quality features
1. Student sample size (“sample”). Actual sample size.
2. Measure of academic outcome (“achmeasure”).
Standardized test=1,
Researcher-made test=2,
Teacher-made test=3,
Other=4.
3. Testing sequence (“testseq”).
Unspecified=0,
Pre-post=1,
Post only=2,
Other=3.
4. Pretest equivalency (“preequiv”). Have the initial differences between groups been
accounted for?
Unspecified=0,
Statistical control (ANCOVA, regression)=1,
Random assignment=2,
Statistical control and random assignment=3,
Gain scores=4,
Other=5.
5. Reported reliability of measures (“reliability”).
Unspecified=00,
Actual reliability statistic.
6. Effect size coefficient (“effsize”).
Actual coefficient.
Meta-Analysis of Distance Education
Learning Point Associates 37
7. Statistics used in determining effect size. (“esstats”).
Means=1,
t-value=2,
F-value=3,
Chi-square=4,
Other=5.
8. Weight (“weight”).
One divided by the actual number of findings/hypotheses in the study.
E. Sources of Invalidity
1. Type of Design (“design”).
Quasi-experimental/nonrandomized one group pretest-posttest=1,
Nonrandomized static-group comparison=2,
Nonrandomized pre-post control group=3,
Time series=4,
Randomized posttest-only control group=5,
Randomized pre-post control group=6,
Other=7.
2. History (“history”). Control for specific events occurring between the first and second
measurement in addition to the experimental variable.
Adequately controlled by design=1,
Definite weakness of design=2,
Possible source of concern=3,
Not a relevant factor=4.
3. Maturation (“maturation”). Control for processes within the participants operating as a
function of the passage of time.
Are there processes within participants operating as a function of the passage of time,
such as growing older or more tired, that might account for changes in the dependent
measure?
Adequately controlled by design=1,
Definite weakness of design=2,
Possible source of concern=3,
Not a relevant factor=4.
4. Testing (“testing”). Control for the effect of taking a test upon the scores of a second
testing.
Adequately controlled by design=1,
Definite weakness of design=2,
Possible source of concern=3,
Not a relevant factor=4.
Meta-Analysis of Distance Education
Learning Point Associates 38
5. Instrumentation (“instrument”). Control for changes in calibration or observers' scores
that produce changes in the obtained measurement.
Adequately controlled by design=1,
Definite weakness of design=2,
Possible source of concern=3,
Not a relevant factor=4.
6. Statistical Regression (“regression”). Control for group selection based on their extreme
scores.
Adequately controlled by design=1,
Definite weakness of design=2,
Possible source of concern=3,
Not a relevant factor=4.
7. Selection Bias (“selection”). Control for biases resulting in the differential selection of
comparison groups.
Adequately controlled by design=1,
Definite weakness of design=2,
Possible source of concern=3,
Not a relevant factor=4.
8. Mortality (“mortality”). Control for differential loss of participants from the
experimental and control groups.
Adequately controlled by design=1,
Definite weakness of design=2,
Possible source of concern=3,
Not a relevant factor=4.
9. Selection-Maturation Interaction (“selectmatur”). Control for interaction between
extraneous factors such as history, maturation, or testing and the specific selection
differences that distinguish the experimental and control groups.
Adequately controlled by design=1,
Definite weakness of design=2,
Possible source of concern=3,
Not a relevant factor=4.
10. Reactive or Interaction Effect of Testing (“testeff”). Control for the influence of pretesting
on the participants' responsiveness to the experimental variable, making the
results for a pre-tested population unrepresentative of the effects of the experimental
variable for the unpre-tested universe from which the participants were selected.
Adequately controlled by design=1,
Definite weakness of design=2,
Possible source of concern=3,
Not a relevant factor=4.
Meta-Analysis of Distance Education
Learning Point Associates 39