• Anymir Orellana, Program Professor, Instructional Technology and Distance Education, Fischler School of Education
and Human Services, Nova Southeastern University, 1750 NE 167th St., North Miami Beach, FL 33162. Telephone: (954)
262-8797. E-mail: [emailprotected]
The Quarterly Review of Distance Education, Volume 7(3), 2006, pp. 229–248 ISSN 1528-3518
Copyright © 2006 Information Age Publishing, Inc. All rights of reproduction in any form reserved.
CLASS SIZE AND INTERACTION
IN ONLINE COURSES
Anymir Orellana
Nova Southeastern University
This article presents findings of a study conducted to determine instructors’ perceptions of optimal class sizes
for online courses with different levels of interaction. Implications for research and practice are also pre-
sented. A Web-based survey method was employed. Online courses studied were those taught sometime in the
last 5 years by a single instructor in undergraduate or graduate programs from U.S. higher education institu-
tions. Instructors described the level of interactive qualities in their most recently taught online course using
a Web version of Roblyer and Wiencke’s (2004) Rubric for Assessing Interactive Qualities in Distance
Courses, and they indicated optimal class sizes according to such qualities. Responses from 131 instructors
were analyzed. On average (a) instructors described their online courses as highly interactive, (b) the actual
class size of the online courses was 22.8, (c) a class size of 18.9 was perceived as optimal to better achieve the
course’s actual level of interaction, and (d) a class size of 15.9 was perceived as optimal to achieve the highest
level of interaction. No relationship was found between online courses’ actual class sizes and their actual level
of interaction.
Modern distance education is a means for
higher education institutions to increase enroll-
ments and students’ access to learning (Lewis,
Alexander, & Farris, 1997). Between 1997 and
2001, the percentage of American higher edu-
cation institutions that offered distance educa-
tion courses increased from 34 to 56, and
course enrollments increased from 1.7 million
to 3.1 million (Wirt, Choy, Rooney, Provasnik,
Sen, & Tobin, 2004). Institutions also seek to
implement quality distance education that
often translates into high initial fixed costs and
variable costs related to delivery of instruction
(Bates, 2000; Bates & Poole, 2003; Morgan,
2000). These variable costs depend on course
enrollments and, hence, class sizes.
Setting class-size limits is a budget-related
matter for administrators (Parker, 2003; Tho-
mas, 1984). Administrators are faced with the
issue of determining an optimal class size to
balance the cost-benefit relationship, while
maintaining manageable faculty workloads
and ensuring quality education. Administrators
often believe that the number of students can
230 The Quarterly Review of Distance Education Vol. 7, No. 3, 2006
be as large as hundreds because there is no
physical space limitation in distance education
(Simonson, 2004). Conversely, in a report of a
year-long faculty seminar (University of Illi-
nois, 1999), the following was concluded:
Because high quality online teaching is
time and labor intensive, it is not likely to
be the income source envisioned by some
administrators. Teaching the same number
of students online at the same level of qual-
ity as in the classroom requires more time
and money. (p. 2)
Class size research is important to educa-
tional policy development. Despite the growth
of distance higher education, little research has
been reported regarding class sizes for online
courses (Boettcher & Conrad, 2004; Parker,
2003; Simonson, 2004). Simonson (2004) sug-
gested that claims of “smaller is better [or that]
it really makes no difference how many, if the
course is organized correctly” (p. 56) are
“myths” of distance education. Most of the
class sizes recommended in the literature for
distance education are based on anecdotal evi-
dence (Simonson, 2004).
In this study, the online class-size problem
was approached from the perspective of the
instructor. It was assumed that different online
courses may have different interactive quali-
ties. Hence, the concern was not to determine a
“one-size-fits-all” optimal class size for online
courses, but to determine optimal class sizes
according to the interactive qualities present in
online courses. For the purpose of the study,
interaction was defined as “a created environ-
ment in which both social and instructional
messages are exchanged among the entities in
the course and in which messages are both car-
ried and influenced by the activities and tech-
nology resources being employed ” (Roblyer
& Wiencke, 2003, p. 81). Interaction is
achieved “through a complex interplay of
social, instructional, and technological vari-
ables” (p. 1).
The purpose of this study was to determine
instructors’ perceptions of optimal class sizes
for online courses with different levels of inter-
action. The level of interaction was measured
with Roblyer and Wiencke’s (2004) Rubric for
Assessing Interactive Qualities in Distance
Courses (RAIQ). The RAIQ is a validated
instrument that measures interactive qualities
through five observable indicators (Roblyer &
Wiencke, 2004): (a) social rapport-building
designs for interaction, (b) instructional
designs for interaction, (c) interactivity of tech-
nology resources, (d) evidence of learner
engagement, and (e) evidence of instructor
engagement. The RAIQ was not used in the
study as a means to imply that the highest lev-
els of interaction were optimal, needed, or
desired in an online course. As Moore and
Kearsley (2005) suggested, the RAIQ was used
in the study as a “means of thinking about what
kind of interaction you [the instructor] want to
facilitate for different types of students and dif-
ferent subject areas” (pp. 145-146).
Online courses studied were those that
(a) counted for credit toward a degree in a
bachelor’s, master’s, or doctoral program from
an American higher education institution;
(b) were taught at a distance at least 80% of the
time using interactive telecommunications
systems, perhaps with occasional traditional
face-to-face activities; and (c) were taught by
one instructor with no teaching assistant, or the
like, sometime in the past 5 years. Class size
was defined as the number of students main-
tained during instruction after the drop period.
Class size did not necessarily reflect the num-
ber of initially enrolled students, or the limit
set by the institution.
The study employed a Web-based survey
research method. Instructors were asked to
determine the level of interactive qualities in
their most recently taught online course using
a Web version of the RAIQ. Instructors were
then asked to indicate what they perceived as
optimal class sizes to better achieve the
course’s actual level of interaction and to bet-
ter achieve the highest possible level of inter-
action, as measured by the RAIQ. Qualitative
comments were also collected from instruc-
tors.
Class Size and Interaction in Online Courses 231
It was anticipated that findings would be
useful as an initial approach to the class size
problem in the field of distance education, spe-
cifically for online courses in higher educa-
tion. It was also anticipated that results might
be applicable to policy development regarding
class-size limits for online courses. The impor-
tance given to interaction in the research, in
best-practice guidelines, and in accreditation
standards for online education served as the
main framework for the study.
REVIEW OF LITERATURE
Research on class size in traditional education
has been conducted for more than a century
(Achilles, 1999). Research in elementary edu-
cation has demonstrated that smaller classes
allow for better student-teacher interaction
(Achilles, 1999; Laine & Ward, 2000; Prit-
chard, 1999). More than 20 states in the United
States have developed and implemented state-
wide policies that limit class sizes in public
schools (Pritchard, 1999). On the other hand,
class sizes in higher education usually can be
as large as the institution deems necessary.
According to Borden and Burton (1999), most
studies focused on higher education have
reported mixed results. Class size mostly
affects what goes on in the classroom and not
student achievement, per se (Gilbert, 1995;
Hanco*ck, 1996; Pascarella & Terenzini, 1991;
Raimondo, Esposito, & Gershenberg, 1990;
Toth & Montagna, 2002).
Gilbert (1995) advocated for large classes
in higher education where group collaboration
is best done. According to Gilbert, “Instruction
which is intimate, interactive and investigative
produces the most positive educational out-
comes. The importance of interaction, partici-
pation and involvement of student learning are
widely recognized … and are, in fact, a part of
effective large class instruction” (p. 5). On the
other hand, Gilbert also suggested that quality
instructor-student interaction is perhaps best
achieved in smaller classes. Brown (as cited in
Pascarella & Terenzini, 1991) and Smith and
Malec (as cited in Pascarella & Terenzini,
1991) found that students’ experiences in large
classes negatively impacted student-faculty
interaction. Also, Pascarella and Terenzini
concluded that evidence suggested that smaller
classes are better than larger ones if the goals
of instruction are “motivational, attitudinal, or
higher-level cognitive processes” (p. 87).
The question as to whether smaller classes
are more conducive for learning than large
ones is also important in distance education.
Instructors also believe that quality of online
instruction is questionable for large class sizes
(Olson, as cited in Olson, 2002; Parker, 2003;
University of Illinois, 1999). Sugrue, Rietz,
and Hasen (1999) conducted a study across
three learning sites to determine relationships
among class size, instructor location, student
perceptions, and performance. Two classes
were taught at a distance via two-way video
and differed in class size and the third class
was taught face-to-face with 36 students.
Results indicated that performance in the two
smaller classes was better than in the large
class. The authors concluded that, without con-
sidering individual differences among learn-
ers, class size influenced performance more
than location did. Also, the authors indicated
that small classes must be kept for successful
multisite distance learning with two-way
video. However, it was not clear to them what
the optimum class size was.
Due to perceived higher demands of stu-
dent-teacher interaction in online courses,
many (e.g., Ko & Rossen, 2004; Sellani &
Harrington, 2002; University of Illinois, 1999)
have considered that instructors’ workload
increases with class size. In a descriptive study
conducted by Berge and Muilenburg (2001),
faculty time and workload were reported as
main barriers for the adoption of online
courses at any stage of the institution’s matu-
rity in implementing distance education.
Instructors’ perceptions of more work in
online courses might be due to the instructor’s
unfamiliarity with the use of the media
(Anderson, 2003; Hislop & Ellis, 2004).
Accordingly, Simonson (2004) called the
232 The Quarterly Review of Distance Education Vol. 7, No. 3, 2006
instructor-perceived-more-time issue the
“‘more work’ myth” (p. 56) that is claimed
among distance education practitioners. This
group usually advocates for smaller classes.
However, small classes might not be appropri-
ate for course designs with emphasis on col-
laborative or group learning activities (Bates &
Poole, 2003; Ko & Rossen, 2004; Vrasidas &
McIsaac, 1999).
Survey research conducted by the National
Education Association (NEA, 2000) showed
that instructors perceived that time, or effort, is
greater when teaching an online course, as
opposed to a face-to-face course. However, the
NEA report also showed that class size was not
related to the amount of online teaching time
estimated by surveyed faculty members.
DiBiase (2000) concluded that the normalized
teaching time per student in the online course
was not greater than in the traditional version.
Similarly, Hislop and Ellis (2003) found no
significant difference in the total time spent by
instructors teaching online versus face-to-face
when time was normalized for class size.
Visser (2000) conducted an experimental case-
study to analyze the time to develop and teach
the graduate-level distance course compared to
a similar traditional course. Time was adjusted
for class size. Visser concluded that online
courses do seem to take more teaching and
development time than the traditional course,
but also noted that delivery time and effort
may depend on the instructor experience and
the level of institutional support.
Determining an optimal class size depends
on multiple factors. According to Bates
(2000), the driving factor that determines the
ideal class size for an online course is the
“amount and nature of the interaction between
the tutor and students [and] student-teacher
ratio is as much determined by educational
philosophy, course design, and student num-
bers as by technology” (p. 129). In addition, a
considerable body of literature presents sets of
best practices and guidelines for course
designs and for interactive strategies that pro-
mote quality distance education. Online strate-
gies range from collaborative group activities,
where interaction among students is essential,
to activities in which more individualized
instructor-student interaction is needed. Addi-
tionally, conventional wisdom suggests that
large class sizes for online courses impact the
amount of individual instructor-student inter-
action (Simonson, 2004). On the other hand,
small class sizes negatively affect interaction
in online community building (Vrasidas &
McIsaac, 1999).
The importance of interaction in the design
of distance courses is also highlighted in
accreditation standards of the Southern Asso-
ciation of Colleges and Schools (2000) and the
Western Cooperative for Educational Tele-
communications (WCET, 2000). Accredita-
tion is the means by which American higher
education institutions are reviewed for quality
(Council for Higher Education Accreditation,
2001) and recommended accreditation stan-
dards should be taken into account in the
development of distance education policies
(Simonson, Smaldino, Albright, & Zvacek,
2003). The Accrediting Commission of Career
Schools and Colleges of Technology (2004)
developed standards of accreditation that “sets
forth the criteria under which the Commission
will recognize programs or courses of study
offered via distance education” (p. 29). Class
size and interaction were addressed under the
following faculty-related standards:
The school ensures that faculty and stu-
dents interact, and provides adequate
means for such interaction
The school must have developed policies
addressing teaching load, class size, time
needed for course development, and the
sharing of instructional responsibilities
which allow for effective teaching using
distance education methods. (p. 29)
The American Association of University
Professors (AAUP, n.d.) has posted sugges-
tions and guidelines for a sample language for
distance education institutional policies and
contract language. The AAUP recommended
the following language for policies concerning
faculty workload and teaching responsibilities:
“Determination of class size for a distance
Class Size and Interaction in Online Courses 233
education class should be based on pedagogi-
cal considerations. Large sections should be
compensated by additional credit in load
assignment in the same manner as traditional
classes” (Workload/Teaching Responsibility
section, ¶ 1). This recommendation is based on
anecdotal evidence:
In the absence of more definitive data,
workload provisions should take into
account the anecdotal evidence that dis-
tance education course development is tak-
ing two to three times as long as
comparable courses taught in the tradi-
tional manner. The same evidence suggests
that the investment of faculty time involved
in teaching a distance education course is
substantially greater than that required for a
comparable traditional course. The time
spent online answering student inquiries is
reported as being more than double the
amount of time required in interacting with
students in comparable traditional classes.
(Workload/Teaching Responsibility sec-
tion, ¶ 1)
In summary, research findings, practical
guidelines and standards, and anecdotal evi-
dence suggest that interaction is affected by
class size. Determining an optimal class size
for an online course is complex and depends
on several factors. Instructors involved in the
design, delivery, and administration of courses
are key elements to successful distance educa-
tion and their perceptions of optimal class
sizes would be useful information to policy
makers. A goal of this study was to determine
such perceptions as they relate to interaction in
online courses.
THEORETICAL FRAMEWORK
As in traditional classrooms, interaction is
considered necessary and desirable for suc-
cessful online learning (Bates, 2000; Fulford
& Zhang, 1993; Lock, 2002; Moore, as cited in
Gresh & Mrozowski, 2000; Offir, as cited in
Gresh & Mrozowski; Roblyer & Wiencke,
2003; Sorensen & Baylen, 2000). Conse-
quently, a model that captures the essence of
theoretical and practical fundamentals of inter-
action is useful. In this respect, Roblyer and
Wiencke (2004) developed and validated a
RAIQ. The model is based on findings from
theory and research related to interaction in
distance education (e.g., Moore, 1989; Wag-
ner, 1994; Yacci, 2000). Roblyer and
Wiencke’s (2004) RAIQ served as the main
framework for this study. According to
Roblyer and Wiencke, the rubric can be used
by instructors as a “tool to allow more mean-
ingful examination of the role of interaction in
enhancing achievement and student satisfac-
tion in distance learning courses” (p. 77). As
Roblyer and Wiencke pointed out, the RAIQ
might help the “design and research of optimal
distance learning environments by helping to
define and quantify observed interaction and
allow empirical assessment of its contribution
to course effectiveness” (p. 95).
METHOD
The study examined the following questions:
What are instructors’ perceptions of optimal
class sizes for online courses with different
levels of interactive qualities? What are typical
class sizes of online courses? What are typical
levels of interactive qualities in online
courses? A Web-based survey research
method was employed. The Class Size and
Interaction Questionnaire (CSIQ) was the
Web-based instrument used for data collec-
tion.
Participants
According to Fowler (1993), “people who
have particular interest in the subject matter
or the research itself are more likely to return
mail questionnaires than those who are less
interested” (p. 4). Hence, in addition to fac-
ulty members who teach college-level online
courses, groups of researchers in the field of
distance education were also considered as
potential participants. Participants were
instructors who, sometime in the past 5 years,
234 The Quarterly Review of Distance Education Vol. 7, No. 3, 2006
had taught an online course as defined in the
study and were sampled from five groups of
interest: (a) presenters of distance education-
related topics at the 2004 National Conven-
tion of the Association for Educational Com-
munications and Technology, (b) researchers
who have published in the journal Quarterly
Review of Distance Education, (c) researchers
who have published in the journal Distance
Learning, (d) researchers who have published
in the American Journal of Distance Educa-
tion, and (e) faculty members of U.S. higher
education institutions that offer online
courses.
Procedures
The Web-based software Surveyor was
used to construct and administer the CSIQ via
the Internet. Invitations and follow-ups to par-
ticipants were also administered by Surveyor.
Confidentiality, anonymity, and one-time
responses were guaranteed by means of a
secure Web-server, automated invitation and
follow-up to participants, and randomly-gener-
ated-password access to the CSIQ. A multi-
stage clustering was conducted to compile a
list of 659 e-mail addresses from the five
groups of interest based on the professional
profile that was published on the selected jour-
nals or posted on the Web. The initial e-mailed
invitation for participation in the research used
Surveyor’s features for survey invitation.
Thirty-four messages were automatically
returned to the researcher because of invalid
e-mail addresses. These 34 addresses were
deleted from the invitation list. Hence, a total
of 625 composed the final list of invitation
recipients.
After receiving the invitation, participants
had 2 weeks to visit the URL that granted
access to the CSIQ. Participants had to use the
unique password randomly generated by Sur-
veyor to access the CSIQ. To reduce the nonre-
sponse rate, a follow-up e-mail was sent to
nonrespondents as a reminder to complete the
CSIQ. Surveyor automatically e-mailed the
invitation letter to those who had not replied 1
week after the initial invitation. Eighty-six
individuals submitted answers to the CSIQ
before the follow-up reminder, and 68 more
after the reminder. A total of 154 responses
were collected. The response rate was 33.8%.
The response rate was computed considering a
total of 625 actual invitation-recipients and
211 replies to the invitation (i.e., 154 actual
respondents to the CSIQ and 57 self-reported
unqualified individuals).
According to Fowler (1993), “The effect of
nonresponse survey estimates depends on the
percentage not responding and the extent to
which those that not responded are biased—
that is, systematically different from the
whole population” (p. 40). To maintain a non-
biased nonresponse rate, several aspects were
considered: (a) sampled individuals were
selected based on their professional profile
(i.e., instructors or faculty members of col-
lege-level online courses), (b) individuals who
did not meet the inclusion criteria were
expected to reply to the e-mailed invitation
and follow-up messages, (c) a conditional
question in the CSIQ automatically directed
respondents to the rest of the CSIQ questions
only if they met the inclusion criteria, and (d)
five nonrespondents were contacted by tele-
phone to determine why they did not respond
to the CSIQ.
From the five nonrespondents who were
telephoned, two indicated that they usually do
not take the time to answer online surveys.
One did not read the e-mailed invitation or
reminder, but indicated that he usually sup-
ported this kind of research and would have
been pleased to participate. Another indicated
that she did not teach online courses. The last
nonrespondent telephoned indicated that she
did not believe that the research problem was
worthwhile or appropriate, and was not willing
to participate.
Instruments
The CSIQ was designed following guide-
lines recommended by Gall, Gall, and Borg
(2003) and by Schonlau, Fricker, and Elliot
Class Size and Interaction in Online Courses 235
(2001) for Web-based questionnaires. The
questionnaire consisted of an initial question to
verify that the respondent met the inclusion
criteria (i.e., sometime in the last 5 years, he or
she had taught an online course as defined in
the study) and four main parts: demographics,
general questions related to the instructor’s
most recently taught online course, Web ver-
sion of the RAIQ, and optimal class-size ques-
tions and comments
Demographics
Questions were formulated to collect
respondents’ age, gender, highest academic
degree, number of years since degree was
awarded, number of years teaching in higher
education, academic rank in faculty position,
general area of teaching from the United
Nations Educational Scientific and Cultural
Organization’s (UNESCO, 1997) Web site,
level of expertise in online teaching on a scale
from 1 (novice) to 5 (very experienced), num-
ber of years teaching online courses, and num-
ber of online courses taught. Respondents also
indicated whether they had received formal
training in online teaching methods.
General questions related to the instruc-
tor’s most recently taught online course. Ques-
tions were formulated to collect the course’s
actual class size, academic level of the pro-
gram (bachelor’s, master’s, or doctoral), dura-
tion in weeks, and semester credits. Questions
were formulated to collect the number of
credit-bearing courses that the instructor
taught during the same academic term, the
Carnegie classification from the Carnegie
Foundation for the Advancement of Teaching
(2005), and type of the institution that offered
the course (public, private for-profit, private
nonprofit).
Web Version of the RAIQ
Roblyer and Wiencke’s (2004) RAIQ was
used in its complete original form, but with a
different layout format suited for the Web.
Specifically, the five elements or indicators for
interactive qualities in a distance course were
separately displayed, as opposed to the origi-
nal matrix-like display. Following is a brief
description of each element:
1. Social rapport-building designs for inter-
action. This element is measured by the
strategies designed for social interaction
among participants. The instructor has
control of the strategies during the design
and implementation phases of instruction.
2. Instructional designs for interaction. This
element is measured by the activities
“designed to encourage, support, and
even require interaction [among partici-
pants]” (p. 87). The instructor has control
of the activities during the design and
implementation phases of instruction.
3. Interactivity of technology resources.
This element is measured by the various
levels of interactivity that are offered by
various technologies. The technologies
“become meaningful components to pro-
mote interaction only in the context of
course designs that make effective use of
them” (p. 88).
4. Evidence of learner engagement. This
element is measured by “the number of
students who reply and who initiate mes-
sages on a frequent basis; send messages
both when required and spontaneously;
and send detailed, informative, well-
developed communications that are
responsive to discussion purposes”
(p. 89).
5. Evidence of instructor engagement. Mea-
sured by the “consistent, timely, and use-
ful feedback to students [from the
instructor]” (p. 89).
Optimal Class-Size Questions and
Comments
Two open-ended questions were formulated
to collect instructors’ perceptions of (a) an
optimal class size that allows for the actual
level of interaction in their most recently
taught online course, and (b) an optimal class
236 The Quarterly Review of Distance Education Vol. 7, No. 3, 2006
size that allows for the highest level of interac-
tion in the RAIQ (i.e., a maximum score of
25). Table 1 presents the interactive qualities
that characterize a course with the highest
level of interaction in the RAIQ. An open-
ended question was formulated to collect par-
ticipants’ comments that they believed would
contribute to the study.
Data Analysis
Data collected from Surveyor were input to
a spreadsheet. The spreadsheet data were then
input to SPSS Student version 7 for Windows
to obtain descriptive statistics. Following is a
description of how the data were organized
and analyzed:
1. Determining levels of interactive qualities
in the RAIQ. The overall level of a
course’s interactive qualities can be low,
moderate, or high (Roblyer & Wiencke,
2004). To obtain the course’s interactive
level, points were assigned to each level-
option under each of the five elements.
There were five options of levels under
each element: low, minimum, moderate,
above-average, and high. Low interactive
qualities were worth 1 point; minimum
interactive qualities were worth 2 points;
moderate interactive qualities were worth
3 points; above-average interactive quali-
ties were worth 4 points; and high interac-
tive qualities were worth 5 points.
Participants could only select one level
per element. The five resulting scores
(i.e., one per element) were totaled, and
according to the interval where the total
fell, the course had one of three interac-
tive levels: low (1 to 9 points), moderate
(10 to 17 points), or high (18 to 25
points). This calculation was done for
each entry in the spreadsheet (i.e., for
each online course described by respon-
dent) and saved as the course’s level of
interactive qualities.
2. Determining class sizes. Descriptive sta-
tistics were obtained for class sizes of
respondents’ most recently taught online
courses. Class-size statistics were
grouped according to (a) the course’s
level of interactive qualities, (b) academic
level of the online course’s program, (c)
type of institution that offered the course,
TABLE 1Highest Levels of Interactive Qualities in a Distance Course in the
Rubric for Assessing Interactive Qualities in Distance Courses (RAIQ)
Element in the RAIQ Description
1. Social/rapport-build-
ing designs for inter-
action
In addition to providing for exchanges of personal information and encouraging student-
student and instructor-student interaction, the instructor provides ongoing course structures
designed to promote social rapport among students and instructor.
2. Instructional designs
for interaction
In addition to the requiring students to communicate with the instructor, instructional activities
require students to develop products by working together cooperatively (e.g., in pairs or small
groups) and share results and feedback with other groups in the class.
3. Interactivity of tech-
nology resources
In addition to technologies to allow two-way exchanges of text information, visual
technologies such as two-way video or videoconferencing technologies allow synchronous
voice & visual communications between instructor and students and among students.
4. Evidence of learner
engagement
By end of course, all or nearly all students (90-100%) are both replying to and initiating
messages, both when required and voluntarily; messages are detailed, responsive to topics, and
are well-developed communications.
5. Evidence of instruc-
tor engagement
Instructor responds to all student queries; responses are always prompt, that is, within 24
hours; feedback always offers detailed analysis of student work and suggestions for
improvement, along with additional hints and information to supplement learning.
Source: Roblyer and Wiencke (2004). Copyright 2004 by M. D. Roblyer. Adapted with permission.
Class Size and Interaction in Online Courses 237
and (d) Carnegie classification of the
institution that offered the course.
3. Determining perceived optimal class
sizes. Respondents’ perceptions of opti-
mal class sizes were grouped according to
levels of interactive qualities previously
calculated, and to the highest possible
level in the RAIQ. Hence, four possible
data groups of perceived optimal class
sizes resulted according to the course’s
level of interactive qualities. Descriptive
statistics were obtained for each group of
data. Subgroups were analyzed and
descriptive statistics were obtained
according to (a) the course’s level of
interactive qualities, (b) academic level of
the online course’s program, (c) type of
institution that offered the course, and (d)
Carnegie classification of the institution
that offered the course.
DISCUSSION OF RESULTS
From 154 CSIQ response-cases to the CSIQ,
23 were not analyzed. The reasons for remov-
ing the 23 cases were as follows: (a) 5 respon-
dents provided a negative answer to the initial
question of the CSIQ, indicating that they did
not meet the inclusion criteria (e.g., they had
teacher assistants, they had not taught in an
American institution, or the face-to-face com-
ponent of the online course was greater than
20%); (b) 17 respondents gave an affirmative
answer to the initial question of the CSIQ, but
did not answer the rest of the questions; (c) 1
respondent indicated a class size of 100, and
the corresponding answers were removed
because they were considered outliers. There-
fore, the final sample was 131 (N = 131).
From 131 respondents, most (61.8%) were
female, had doctoral degrees (82.4%), taught
in the area of education (47.3%), on average
perceived themselves as very experienced in
online teaching (4.2 over 5), and had received
formal training in online teaching (52.7%).
Most of respondents’ online courses were
taught in public (71.8%), doctoral-research
universities (68.7%), and in graduate programs
(53.4% master’s and 17.6% doctoral).
Following is a discussion of results related
to the study’s research questions. Results were
interpreted bearing in mind demographics of
respondents, the type of online courses stud-
ied, and the scope and purpose of the RAIQ
and of the CSIQ.
What Are Typical Class Sizes of Online
Courses?
Results from the CSIQ indicated that actual
class sizes (CS) for the 131 respondents ranged
from 4 to 81. The mode was 20 and the average
was 22.8. Almost 62% of respondents’ courses
had 20 or fewer students, and only 2 courses
had a CS greater than 65. From the results, it
can be concluded that for online courses, as
defined in the study, the average CS was
approximately 23, the most frequent CS for an
online course was 20, and most courses
(61.8%) had a CS smaller than or equal to 20.
According to data posted in U.S. News &
World Report (“E-learning,” 2005), accredited
higher education institutions that offer online
graduate-programs in education have reported
class size limits of 23, on average. Even
though the CSIQ did not examine the accredi-
tation status of the institution, the average CS
identified by the CSIQ is consistent with the
data posted in U.S. News & World Report. On
the other hand, the NEA (2000) reported that
one third of online courses had 20 or fewer stu-
dents, and two thirds had 21 to 40. Similarly,
according to the Higher Education and Policy
Council of the American Federation of Teach-
ers (2000), only one third of instructors taught
online courses with 20 or fewer students. In
contrast, results of this study indicated that
most respondents (61.8%) reported a class size
of 20 or less, and only a 27.8% reported a class
size from 21 to 40. It seems that more recent
courses, taught during the years 2000 and
2005, are smaller than those taught before the
year of publication of the NEA report. How-
ever, the specific characteristics of the online
courses studied (see Table 2) and the limita-
238 The Quarterly Review of Distance Education Vol. 7, No. 3, 2006
tions of the sample of participants prevent
making a generalization to other distance
courses. Therefore, it is not appropriate to
draw conclusions about typical class sizes of
online courses from comparing these studies.
Table 3 presents descriptive statistics for
class sizes. A more in-depth analysis indicated
that, in public doctoral-research universities,
the largest average class size resulted for
courses in bachelor’s programs (43.5), and the
smallest average class size resulted for courses
in doctoral programs (15). These results were
to be expected. Public institutions usually have
higher enrollments than private institutions,
and bachelor’s programs usually enroll more
students than doctoral programs. Hence, class
TABLE 2Characteristics of Online Courses According to Respondents
to the Class Size and Interaction Questionnaire (N = 131)
Measure Min Max Average Standard Deviation
Actual class size 4 81 22.8 13.7
Number of weeks 4 20 14.2 2.8
Interactive level* 9 25 18.8 3.8
Semester credits 1 6 3.2 0.7
Note: Min = Smallest score reported, Max = Largest score reported.
*Interactive level = Sum of points of the five elements of interactive qualities described in the questionnaire; low interactive level = 1 to 9
points, moderate interactive level = 10 to 17 points, high interactive level = 18 to 25 points.
TABLE 3Descriptive Statistics for Class Sizes of Online Courses According to Respondents to the Class Size and
Interaction Questionnaire (N = 131)
Classification Min. Max. Average
Standard
Deviation n
Carnegie Classification of Institution
Doctoral 4 81 24.7 15.4 90
Master’s 4 37 19.5 7.4 29
Other 8 35 17.3 7.7 12
Type of Institution
Public 4 81 24.4 15.1 94
Private for-profit 8 35 20.3 7.7 10
Private non-profit 7 45 18.4 8.8 24
Other 13 23 17.0 5.3 3
Academic Level of Online Courses
Bachelor’s 7 81 31.5 18.3 38
Master’s 4 55 19.7 9.7 70
Doctoral 7 35 18.0 7.8 23
Interactive Level of Online Courses*
Low 8 20 14.0 8.5 2
Moderate 7 81 25.8 15.9 43
High 4 65 21.6 12.4 86
Note: Min = Smallest score reported, Max = Largest score reported.
*Interactive level = Sum of points of the five elements of interactive qualities described in the questionnaire; low interactive level = 1 to 9
points, moderate interactive level = 10 to 17 points, high interactive level = 18 to 25 points.
Class Size and Interaction in Online Courses 239
sizes were expected to be larger for online
courses in bachelor’s programs, and for
courses in public institutions.
What Are Typical Levels of Interactive
Qualities in Online Courses?
It was assumed that online courses may
have different interactive qualities and, hence,
different interactive levels (IL), as measured
by the RAIQ. Results from the CSIQ showed
that most respondents (65.6%) perceived that
their online course had a high IL, 32.8% a
moderate IL, and only a 1.5% a low IL. On
average, respondents perceived that their
online courses had a high-interactive level
(18.8 over 25 possible points). Specifically,
the online courses studied could be character-
ized as having above-average levels of social/
rapport-building designs for interaction, of
instructional designs for interaction, of evi-
dence of learner engagement, and of evidence
of instructor engagement. On the other hand,
these online courses could be characterized as
having a moderate level of interactivity and of
technology resources. The standard deviation
of interactive levels was 3.8. From these
results, it can be concluded that almost all
online courses (98.5%), that were taught dur-
ing the years of 2000 and 2005, were moder-
ately to highly interactive without much
variability in their interactive qualities, as
measured by the RAIQ.
Some respondents to the CSIQ commented
that the RAIQ might not be an appropriate
instrument to measure interaction in online
courses. Moreover, respondents who com-
mented about the interactive level in online
courses indicated that the highest levels, as
measured by the RAIQ, are not necessarily
needed, feasible, or desirable. Some indicated
that a high level of interaction did not neces-
sarily require synchronous communication,
video technologies, or such a demanding
instructor engagement as described for the
highest level of the RAIQ (e.g., 24 hours turn-
around response time and instructor’s detailed
responses to every student query). As previ-
ously mentioned, it was not implied in this
study that the highest interactive level was
needed or desirable in an online course. The
purpose of the study was to use the RAIQ to
determine interactive levels of online courses
and obtain information about class sizes
according to these levels.
Most respondents described their online
course as moderately and highly interactive.
Also, results indicated no statistical relation-
ship between CS and IL (see Table 4). The lat-
ter might indicate that CS does not seem to
have an effect on the course’s interactive qual-
ities. Results also indicated that the average CS
(21.6) of highly interactive online courses was
smaller than the average CS (25.8) of moder-
ately interactive ones. Generally speaking,
because it has not been agreed upon in the lit-
erature what actually constitutes a large or a
small online class, it cannot be concluded from
these results that a small CS allows a higher IL
than a large CS, or that highly interactive
online courses have smaller CS than moder-
ately interactive ones. From the results, it can
be concluded that, even though highly interac-
tive online courses that were studied had a
smaller average CS than moderately interac-
tive courses, CS does not seem to be related to
the level of interaction. Respondents com-
mented that other factors, which were also sug-
gested in the literature, might affect
interaction. Some of the mentioned factors
were instructors’ time commitment and work-
load in face-to-face traditional activities (e.g.,
administrative and teaching), course content,
students’ characteristics, and limitations of
technology.
The CSIQ did not measure instructors’
teaching-time commitment or workload in tra-
ditional face-to-face-activities. The CSIQ
measured the number of online courses taught
during the same term (NOCT) including the
online course described. The NOCT did not
measure instructor’s workload completely, but
it was considered to be an indicator of instruc-
tor’s commitment in online teaching during an
academic term. The average NOCT was 2.4
and ranged from 1 to 9. Most respondents’
240 The Quarterly Review of Distance Education Vol. 7, No. 3, 2006
(66.9%) taught, at most, two online courses at
the same time. No relationship was found
between the interactive level and NOCT (see
Table 4).
The CSIQ did not examine characteristics
of students, per se. However, to some extent,
the academic level of the course is related to
the type of students (e.g., students are usually
younger in bachelor’s programs than in gradu-
ate programs). Common wisdom suggests that
graduate courses are more interactive, or
should be more interactive, than undergraduate
courses. However, results from the CSIQ indi-
cated only two online courses with a low inter-
active level, and both were reported at the
master’s academic level. Fifty-nine percent of
the total number of highly interactive courses
was taught in master’s programs, and 21% in
doctoral programs. Bachelor’s online courses
were reported as moderately (55.3%) and
highly (44.7%) interactive. Furthermore, no
relationship was found when analyzing differ-
ences among average scores of interactive lev-
els within groups of courses, per academic
level.
These results indicate that there is not a
strong relationship between the academic level
of the course and the interactive levels of the
studied online courses. Moderate and high
interactive qualities reported for bachelor’s
online courses might be a reflection of younger
students that have embraced technology-medi-
tated courses in different ways than, perhaps
older, graduate students. The assumption that
traditional students at the bachelor’s level are
not as interactive as graduate students might
not be applicable for online undergraduates.
Nonetheless, highly interactive online courses
were more frequent in graduate level programs
than in undergraduate programs.
What Are Instructors’ Perceptions of
Optimal Class Sizes for Online Courses
With Different Levels of Interactive
Qualities?
In distance education, anecdotal class-size
evidence is mostly related to two aspects that
Simonson (2004) denominated “myths of dis-
tance education” (p. 56): (a) It takes more time
TABLE 4Intercorrelations for Selected Measures Examined With the Class Size and Interaction Questionnaire (N = 131)
Measure Age CS YTHE NOCT LE OCS OCSL5 IL NCST FT YTO
Age —
CS −.25** —
YTHE .51** −.12 —
NOCT −.05 .06 .12 —
LE .13 −.03 .17 .34** —
OCS −.19* .79** −.07 .00 .04 —
OCSL5 −.20* .66** −.14 −.02 .08 .81** —
IL .04 −.12 −.13 .13 .25** −.18* −.08 —
NCST −.03 .02 .07 .15 .08 .03 −.02 −.09 —
FT .05 −.16 .06 −.01 .01 −.15 −.10 − .01 −.01 —
YTO .06 .03 .28** .70** .43** .02 − .02 − .14 − .13 .09 —
Note: CS = class size, YTHE = years teaching higher education, NOCT = number of online courses taught, LE = level of
expertise, OCS = optimal class size, OCSL5 = optimal class size for highest interactive levels, IL = interactive level of the
course, NCST = number of online courses taught during the same term, FT = formal training in online teaching, YTO =
years teaching online courses.
*p < 0.05. **p < .01.
Class Size and Interaction in Online Courses 241
to teach online, therefore smaller classes are
needed—the “more-work myth” usually advo-
cated by instructors; and (b) as long as the
course is organized right, it does not matter
how big the class is because there is no physi-
cal space limitation—a myth usually advo-
cated by administrators. Results from this
study seem to support the more-work myth of
distance education.
Respondents indicated that, on average, an
optimal class size (OCS = 18.9) should be
smaller than the actual class size (CS = 22.8).
Results indicated a strong positive correlation
(r = .79) between CS and OCS to support this
conclusion. On the other hand, a very low neg-
ative correlation (r = −.18) between the inter-
active level and OCS seems to indicate that the
higher the interactive level the smaller the
OCS. Hence, it can be concluded that, in gen-
eral, respondents perceived that a smaller OCS
than CS was needed to allow for moderate and
high levels of interactive qualities in their
online courses. Table 5 presents more detailed
descriptive statistics for optimal class sizes for
online courses.
A more detailed analysis of the data
revealed that 23% of respondents believed that
the optimal class size should be greater than
the actual class size. Out of this 23%, 73%
taught courses with an actual class size less
than or equal to 15. Most of these courses
(74%) were perceived as highly interactive.
This might indicate that, for class sizes of less
than or equal to 15, most respondents felt that
more students were necessary to better achieve
the highly interactive qualities present in their
online courses.
Hence, from the results of this study, it can-
not be absolutely determined that higher inter-
active courses, as measured by the RAIQ,
require small classes. These findings might be
an indicative that instructors perceived that
TABLE 5Descriptive Statistics for Optimal Class Sizes for Online Courses According to Respondents to the Class Size
and Interaction Questionnaire (N = 131)
OCS OCSL5
Category Min. Max. M SD Min. Max. M SD n
Interactive Level of Online Courses
Low 15 25 20.0 7.1 6 15 10.5 6.4 2
Moderate 10 80 21.1 15.9 5 40 15.6 6.2 43
High 7 50 17.7 7.6 6 50 16.1 6.9 86
Carnegie Classification of Institution
Doctoral 7 80 19.4 10.1 5 50 16.4 6.9 90
Master’s 8 40 18.2 6.3 8 40 15.1 6.6 29
Other 8 35 16.2 6.8 8 35 13.5 3.9 12
Academic Level of Online Courses
Bachelor’s 10 80 25.3 12.6 5 40 19.3 8.4 38
Master’s 7 50 17.0 5.7 7 50 14.8 5.8 70
Doctoral 7 20 14.0 3.4 8 20 13.5 2.8 23
Type of Institution
Public 7 80 20.2 10.1 5 50 16.6 7.4 94
Private F-profit 10 25 15.9 4.5 100 20 13.9 3.0 10
Private N-profit 8 25 15.5 4.1 8 20 13.9 3.4 24
Other 10 20 15.0 5.0 8 20 14.3 6.0 3
Note: OCS = Perceived optimal class size of online course according to its interactive qualities, OCSL5 = Perceived optimal class size of
online course if it had the highest level of interactive qualities in the questionnaire, Min = Smallest score reported, Max = Largest score
reported, M = Average of scores, SD = Standard deviation of scores.
242 The Quarterly Review of Distance Education Vol. 7, No. 3, 2006
they needed smaller classes than what they
actually had in order to better achieve moder-
ate and high interactive level, but large enough
(e.g., larger than 15) perhaps to increase the
level of interaction in low-interactive courses.
In addition to identifying a smaller average
of optimal class sizes than average actual class
sizes, results indicated that online courses at
the highest interactive levels should have an
average class size of 15.9, which was smaller
than the average optimal class size (18.9).
Also, a strong positive correlation (r = .81)
between optimal class size and optimal class
size for highest interaction was found (see
Table 4). A closer examination of the data
revealed that every respondent perceived that a
smaller class size than the optimal class size
was needed to achieve the highest possible
level of interactive qualities in the RAIQ. The
latter might indicate that respondents per-
ceived that achieving the highest levels of
interaction in the RAIQ might demand from
them more effort per student and, thus, teach-
ing a course with the highest interactive quali-
ties would require a much smaller class size.
Results from this study seem to support the
literature that reports on instructors’ beliefs
that online teaching takes more time or effort
than face-to-face courses. On the other hand,
experimental studies have reported mixed
results about online teaching time or effort.
The literature has also suggested that perhaps
this more-work perception is because of
instructors’ unfamiliarity with technology, or
little experience in online teaching. Perhaps
less-experienced instructors prefer smaller
classes. However, results from this study indi-
cated no relationship between instructors’
level of expertise and both types of perceived
optimal class sizes (i.e., OCS and OCSL5).
The before-mentioned precludes concluding
that teaching experience is related to instruc-
tors’ perceptions of smaller classes to allow for
higher levels of interactive qualities in online
courses.
On the other hand, as seen in Table 4, a very
low negative relationship resulted between
respondents’ age and CS (r = −.25), between
age and OCS (r = −.19), and between age and
OCSL5 (r = −.20). These correlations indicate
that older instructors perhaps prefer smaller
classes than do younger instructors. No statis-
tical relationship was found between respon-
dents’ age and their perceived level of
expertise in online teaching. Also, the number
of years teaching in higher education, the num-
ber of online courses taught, the number of
years teaching online courses, and the level of
expertise were not related to any measure of
class size. In traditional face-to-face settings, it
is customary for department heads to assign
larger classes to new instructors and smaller
classes to instructors with more years teaching
experience. Nonetheless, results of the study
indicated that for online teaching, regardless of
any of the studied indicator of teaching experi-
ence in higher education, instructor’s age was
the factor related to CS. These results might
indicate that, regardless of instructors’ level of
expertise in online teaching, older instructors
taught smaller classes, and preferred smaller
OCS and OCSL5 than younger instructors.
IMPLICATIONS FOR RESEARCH
AND PRACTICE
The theoretical framework for this study was
Roblyer and Wiencke’s (2004) RAIQ, which
was based on several theories of interaction.
Because of the applied nature of the study,
results had implications for practice. Such
implications are mainly related to the decision-
making of class size-related policies that meet
accreditation standards for online programs.
Accreditation is the means by which Amer-
ican higher education institutions are evaluated
for quality. Institutions seek accreditation
through their policies, among which are class
size-related policies. As stated in the literature
review, regional accrediting commissions have
developed a set of guidelines, or quality assur-
ance standards, to reflect current best practices
in electronically offered programs that affect
more than 3,000 colleges and universities in
the United States (CHEA, 2001). The follow-
Class Size and Interaction in Online Courses 243
ing standard exalts the importance of interac-
tion in the design of distance courses and
programs: “The importance of appropriate
interaction (synchronous or asynchronous)
between instructor and students and among
students is reflected in the design of the pro-
gram and its courses, and in the technical facil-
ities and services provided” (WCET, 2000,
p. 8).
From the mentioned standards, a main
aspect that can be related to results of the study
is how appropriate interaction and effective
teaching can be achieved through the design of
online courses. Interaction has been a concept
defined and measured in multiple ways in dif-
ferent practical and theory-based publications.
Hence, the appropriateness of interaction can
be a vague term that may be measured in any
way an institution decides. If the appropriate-
ness of interaction is to be measured by the
RAIQ, and moderate and high interaction were
appropriate levels, then almost all online
courses studied had an appropriate interaction.
However, if the appropriate level is the highest
possible level in the RAIQ, then very few
courses met this standard for quality. More-
over, most respondents commented that the
highest level in the RAIQ was not necessarily
needed, feasible, or desirable.
Hence, two major implications for practice
can be derived from this study: accrediting
organizations might need to clearly indicate
how they expect institutions to measure the
appropriateness of interaction; and the highest
interactive level of the RAIQ is not always an
appropriate level of interaction for an online
course. Inherent to these implications is that a
design of the online course that reflects the
appropriateness of interaction is subject to the
characteristics the course. Once again, if the
RAIQ is to be used to assess the design of the
course through each of its five elements, then
multiple combinations (i.e., scores for each
element in the RAIQ) yield a certain level of
interaction. That is, each element contributes
to determine the level of interaction of the
course. Thus, determining whether the design
reflects an appropriate interaction is a complex
task.
Results from this study, in addition to the
literature about interaction and class size,
could be used by accrediting organizations to
indicate that different levels of interaction can
be appropriate for an online course, and that
different course designs can allow for appro-
priate levels of interaction. Furthermore, the
literature and research does not support that
more interaction in online courses is necessar-
ily more conducive to learning, party because
of the different ways to define and measure
interaction.
Regarding class size and interaction, com-
mon wisdom has held that smaller class sizes
for online courses allow for more interaction.
In their recommended standard for distance
education courses, the AFT (2000) stated that
“class size should encourage a high degree of
interactivity [and that] given the time commit-
ment involved in teaching through distance
education, smaller class sizes should be con-
sidered, particularly at the inception of a new
course” (p. 11). However, experimental
research has not supported that smaller classes
allow for a high level of interactivity. Further-
more, it has not been agreed upon in the litera-
ture what actually constitutes a large or small
online class. In essence, determining what
actually constitutes a large or a small class is a
complex task that does not depend on absolute
criteria, but the perceptions of instructors
might give some insight to approaching the
problem. In this sense, results from this study
could be used to set practical lower and upper
bounds of class sizes for online courses with
moderately interactive and highly interactive
levels.
Other implications for practice are similarly
related to institutional policy-making. Results
indicated no statistical relationship between
actual class size and the interactive level of the
studied online courses, but results did indicate
a low negative correlation between optimal
class size and interactive level. Generally
speaking, respondents seemed to have per-
ceived that they would require more time and
244 The Quarterly Review of Distance Education Vol. 7, No. 3, 2006
commitment if the number of students
increased, when they indicated smaller optimal
class sizes to achieve the highest levels of
interaction in the RAIQ. However, it cannot be
concluded from these results that class size
alone determines the levels of interaction in an
online course.
Respondents to the CSIQ commented that
other factors might determine the level of
interaction in online courses. Some of the men-
tioned factors were instructors’ time commit-
ment, instructors’ workload in face-to-face
traditional activities, and the role of the adjunct
figure as part-time faculty. In this regard, Gell-
man-Danley and Fetzner (1998) considered
that policies related to labor-management
(e.g., class size, assignment of full-time or
adjunct faculty, and workload) were among
the most difficult to develop and included the
toughest questions to ask. Johnstone (2004)
raised the question on whether full-time fac-
ulty members are completely ready to adapt to
online teaching, or whether they were really
the best ones to assist students online. The fig-
ure of readily skillful professionals, as part-
time adjunct faculty, is an alternative for insti-
tutions to fill this possible gap. As Johnstone
pointed out,
One institutional practice that is challenged
by distance learning focuses on who should
be doing the “teaching” [and that] if part-
time faculty members, or adjunct faculty,
are to be the core workforce for online
instruction, then institutions that use a lot
of online teaching may need to develop a
new category of professional employees.
(p. 396)
Nonetheless, adjuncts usually hold other
full-time jobs that prevent them from trying to
reach higher interactive levels in their online
courses, regardless of class size. Respondents
to the CSIQ commented the following: “If I’m
teaching a class (as an adjunct) in addition to
my ‘regular’ full-time job, I may not incorpo-
rate as many interactive activities, regardless
of class size” (Respondent 32); and “most
adjunct professors have other jobs and tend to
do feedback two or three times a week … not
daily” (Respondent 7). Incorporating this kind
of professional workforce, in addition to the
new required roles of full faculty, suggest that
institutions need to develop better ways to
determine teaching workloads that adequately
measures the effort, time-commitment, and
dedication of the instructor in online teaching
tasks, especially interacting with individual
students.
RECOMMENDATIONS FOR
FUTURE RESEARCH
More research in online education is needed to
support or reject the assumption that smaller
class sizes are needed for higher interactive
levels, or even that higher interactive levels are
more conducive to learning than lower interac-
tive levels. Examining the following questions
might support or reject the commonly held
belief that more interaction is better for learn-
ing, and might also help examine whether what
instructors perceived as optimal class size is
better for interaction and learning outcomes: Is
there a relationship between class size and
learning outcomes? Is there a relationship
between the level of interaction, as measured
by the RAIQ, and learning outcomes? Are
there significant differences in levels of inter-
action and learning outcomes among different
online courses with the same perceived opti-
mal class size? Are there significant differ-
ences in levels of interaction and learning
outcomes among similar online courses with
different perceived optimal class size?
On the other hand, respondents’ perception
of smaller optimal class size than actual class
size, on average, might be an indicator that
instructors believed that a larger class size
implies more time commitment and workload.
Nonetheless, class size itself might not be an
aspect that affects online-learning outcomes.
The literature in traditional education has sug-
gested that what happens in the class is what is
actually affected by the class size. Respon-
dents commented that the characteristics of the
online course and of the students, as well as
Class Size and Interaction in Online Courses 245
instructor’s workload, are elements that affect
interaction in an online course. As the IHEP
(2000) suggested for online courses, “Maxi-
mum class size relates more to faculty course
workload than student outcomes. It appears,
therefore, that a specific benchmark for class
size is ill advised, and much more experimen-
tation needs to be conducted” (p. 18).
Hence, additional research questions can be
examined to determine relationships between
instructor’s workload and online class size: Is
there a significant difference between online-
teaching time commitments among online
courses with the same class size and taught by
instructors with different workloads? Is there a
significant difference between online-teaching
time commitments among online courses with
different class sizes and taught by instructors
with similar workloads? How is online teach-
ing time-commitment affected by class size?
How is interactivity affected by the overall
workload of instructors?
Results of this study did not support the
commonly held assumption that graduate stu-
dents interact more than undergraduate stu-
dents. Considering students’ characteristics is
paramount when designing any instruction.
Online instruction poses new challenges to
designers because younger generations of stu-
dents have practically embraced communica-
tions technology as living style. Online
education requires a self-motivated student
capable of using communications technology,
regardless of the program’s academic level.
Both types of students (i.e., graduate and
undergraduate) in online courses that were
offered no more than 5 years ago were perhaps
more technology savvy than students of online
courses that were offered longer ago. Results
indicated a larger average class size for under-
graduate online courses, and that undergradu-
ate online courses were also moderately and
highly interactive, as measured by the RAIQ.
Future research could be conducted to support
or reject the assumption that larger class sizes
are adequate for younger undergraduate stu-
dents because, perhaps, they do not interact as
much as older graduate students.
Some of the before recommended research
issues involve exploring the interactive level
of online courses, as measured by the RAIQ.
Other instruments that have been reported in
the literature can also be used to measure inter-
action. Moreover, respondents to the CSIQ
commented about possible limitations of the
RAIQ to measure interaction. Results of this
study indicated no relationship between actual
class size and interactive levels of the studied
online courses, but perhaps different indicators
of interactive levels would show a relation-
ship. Thus, future research is recommended to
examine the relationship between interaction
and class size as measured by other instru-
ments.
Qualitative research can also contribute to
examine the optimal class-size problem. From
the standpoint of quality in online courses, stu-
dents and instructors might have different per-
spectives of what is an optimal class size. On
the other hand, as respondents to the CSIQ
commented, administrators usually establish
class-size limits and then the instructor must
accommodate the teaching methods accord-
ingly. If results of optimal class size from this
study are taken as benchmarks, a qualitative
study might examine the question: How do
instructors and students behave in similar
online courses with the average optimal class
size?
An assumption that was derived from the
literature is that perhaps less experienced
instructors prefer smaller classes. Results of
this study indicated that regardless of instruc-
tors’ level of expertise in online teaching, older
instructors taught and preferred smaller classes
than did younger instructors. Also, more expe-
rienced instructors seemed to have perceived
their courses as having higher levels of interac-
tive qualities. Some research questions that
arise from these results are related to instruc-
tor’s age: If older instructors perceive them-
selves as having similar levels of experience in
online teaching than younger instructors, why
do they prefer teaching smaller classes? If
older instructors can achieve similar interac-
tive levels in their online courses, why do they
246 The Quarterly Review of Distance Education Vol. 7, No. 3, 2006
prefer teaching smaller classes? Should depart-
ment heads assign larger classes to younger
faculty members?
CONCLUSIONS
Results of this study were intended to be prac-
tical. Optimal class sizes from the perspective
of the instructor were thought to be helpful to
policymakers who are trying to establish class-
size limits for online courses. Limitations of
the study were inherent to the research method
employed (i.e., recruitment of participants,
availability and credibility of respondents, and
limitations of the instruments), and results are
likely to be applicable to online courses as
defined in the study. Future research is recom-
mended to examine class size and interaction
from the perspectives of administrators and of
students.
Findings indicate that, even though the
actual class sizes of the studied online courses
were not related to their actual interactive
qualities and that most respondents perceived
their online courses as moderately and highly
interactive, respondents still believed that they
needed smaller classes to achieve higher inter-
active levels (i.e., an average class size of 22.8
versus a perceived average optimal class size
of 18.9). Furthermore, the data indicate that
every respondent believed that even smaller
class sizes were needed to achieve the highest
interactive level possible in the RAIQ (i.e., an
average of 15.6).
Because interaction is a concept that has
been measured in different ways in research
and practice, accrediting organizations might
need to clearly indicate how an institution is to
measure for appropriate interaction reflected in
the design of the online course in order to meet
quality standards. Also, institutions should
take recommendations from consortia cau-
tiously. Specifically, recommendations of hav-
ing smaller classes to allow for high
interactivity because it has not been supported
by research and it has not been agreed upon
what actually constitutes a large or a small
online class. However, respondents perceived
that smaller classes were needed to achieve the
actual interactive level in their online courses.
This might be because of a perceived increased
effort if they had more students. Hence, for
future research, it is highly recommended to
examine the relationship between class size
and instructors’ workload and between class
size and online teaching time commitment.
REFERENCES
Achilles, C. (1999). Let’s put kids first, finally: Get-
ting class size right. Thousand Oaks, CA: Cor-
win Press.
Accrediting Commission of Career Schools and
Colleges of Technology. (2004). ACCSCT stan-
dards of accreditation. Retrieved January 14,
2005, from http:// www.accsct.org/library/
fullstandards.pdf
American Association of University Professors.
(n.d.). Suggestions and guidelines sample lan-
guage for institutional policies and contract lan-
guage. Retrieved September 9, 2004, from http:
//www.aaup.org/Issues/DistanceEd/Archives/
speccmt/deguide.htm
American Federation of Teachers. (2000). Distance
education: Guidelines for good practice. Wash-
ington, DC: Author. Retrieved April 7, 2004,
from http://www.aft.org/higher_ed/
downloadable/distance.pdf
Anderson, T. (2003). Modes of interaction in dis-
tance education: Recent developments and
research questions. In M. G. Moore & W. G.
Anderson (Eds.), Handbook of distance educa-
tion (pp. 129-144). Mahwah, NJ: Erlbaum.
Bates, A. (2000). Managing technological change.
San Francisco: Jossey-Bass.
Bates, A., & Poole, G. (2003). Effective teaching
with technology in higher education: Founda-
tions for success. San Francisco: Jossey-Bass.
Boettcher, J., & Conrad, R. (2004). Faculty guide
for moving teaching and learning to the Web
(2nd ed.). Phoenix, AZ: League for Innovation.
Borden, V., & Burton, K. (1999, May/June). The
impact of class size on student performance in
introductory courses. Paper presented at the
annual forum of the Association for Institutional
Research, Seattle, WA. (ERIC Document Repro-
duction Service No. ED433782)
Class Size and Interaction in Online Courses 247
Berge, Z., & Muilenburg, L. (2001). Obstacles
faced at various stages of capability regarding
distance education in institutions of higher edu-
cation. TechTrends, 45(4), 40-45. (ERIC Docu-
ment Reproduction Service No. EJ631362)
Carnegie Foundation for the Advancement of
Teaching. (2005). Category definitions.
Retrieved February 2, 2005, from http://
carnegiefoundation.org/Classification/
CIHE2000/defNotes/Definitions.htm
Council for Higher Education Accreditation.
(2001). Fact sheet #2: The role of accreditation
and assuring quality in electronically delivered
distance learning. Retrieved April 29, 2004,
from http://www.chea.org/pdf/
fact_sheet_2_dist_learn_02.pdf
DiBiase, D. (2000). Is distance teaching more work
or less? The American Journal of Distance Edu-
cation, 14(3), 6-20.
E-learning, online graduate degrees: Education
regionally accredited. (2005). Retrieved May
29, 2005, from http://www.usnews.com/usnews/
edu/elearning/tables/edu_reg.htm
Fowler, F. J., Jr. (1993). Survey research methods
(2nd ed.). In L. Brickman & D. J. Rog (Series
Eds.), Applied social research methods series:
Vol. 1. Newbury Park, CA: Sage.
Fulford, C. P., & Zhang, S. (1993). Perceptions of
interaction. The American Journal of Distance
Education, 7(3), 8-21.
Gall, M., Gall, J., & Borg, W. (2003). Educational
research: An introduction (7th ed.). Boston:
Allyn & Bacon.
Gellman-Danley, B., & Fetzner, M. (1998). Asking
the really tough questions: Policy issues for dis-
tance learning. Online Journal of Distance
Learning Administration, 1(1). Retrieved April
29, 2004, from http://www.westga.edu/
~distance/danley11 .html
Gilbert, S. (1995). Quality education: Does class
size matter? Association of Universities and Col-
leges of Canada research file, 1(1). Retrieved
June 2, 2004, from http:// www.aucc.ca/pdf/
english/publications/researchfile/1995-96/
vol1n1_e.pdf
Gresh, K., & Mrozowski, S. (2000, October). Fac-
ulty/student interaction at a distance: Seeking
balance. Paper presented at EDUCAUSE 2000:
Thinking IT Through, Nashville, TN. (ERIC
Document Reproduction Service No.
ED452805)
Hanco*ck, T. (1996). Effects of class size on college
student achievement. College Student Journal,
30, 479-481.
Hislop, G., & Ellis, H. (2004). A study of faculty
effort in online teaching [Electronic version].
Internet and Higher Education, 7(1), 15-31.
Retrieved May 27, 2004, from http://www
.sciencedirect.com
Johnstone, S. M. (2004). A policy perspective on
learning theory and practice in distance educa-
tion. In T. M. Duffy & J. R. Kirkley (Eds.),
Learner-centered theory and practice in dis-
tance education: Cases from higher education
(pp. 395-403). Mahwah, NJ: Erlbaum.
Ko, S., & Rossen, S. (2004). Teaching online: A
practical guide (2nd ed.). Boston: Houghton
Miffin.
Laine, S., & Ward, J. (Eds.). (2000). Using what we
know, a review of the research on implementing
class-size reduction: Initiatives for state and
local policymakers. Oak Brook, IL: North Cen-
tral Regional Educational Laboratory.
Lewis, L., Alexander, D., & Farris, E. (1997). Dis-
tance education in higher education institutions
(NCES 98-062).Washington, DC: U.S. Govern-
ment Printing Office. Retrieved May 19, 2004,
from http://nces.ed.gov/pubs98/98062.pdf
Moore, M. (1989). Three types of interaction. The
American Journal of Distance Education, 3(2),
1-6.
Moore, M., & Kearsley, G. (2005). Distance educa-
tion: A systems view (2nd ed.). Belmont, CA:
Wadsworth.
Morgan, B. (2000). Is distance learning worth it?
Helping to determine the costs of online courses.
(ERIC Document Reproduction Service No.
ED446611)
National Education Association. (2000). A survey of
traditional and distance learning higher educa-
tion members. Washington, DC: Author.
Retrieved April 28, 2004, from http://
www.nea.org/he/abouthe/dlstudy.pdf
Olson, C. (2002). Leadership in online education:
Strategies for effective online administration and
governance. In K. Rudestam & J. Schoenholtz-
Read (Eds.), Handbook of online learning: Inno-
vations in higher education and corporate train-
ing (pp. 237-256). Thousand Oaks, CA: Sage.
Parker, A. (2003). Motivation and incentives for
distance faculty. Online Journal of Distance
Learning Administration, 6(3). Retrieved April
27, 2004, from http://www.westga.edu/
~distance/ojdla/fall63/parker.htm
248 The Quarterly Review of Distance Education Vol. 7, No. 3, 2006
Pascarella, E., & Terenzini, P. (1991). How college
affects students. San Francisco: Jossey-Bass.
Pritchard, I. (1999). Reducing class size: What do
we know? (SAI 98-3097) [Electronic version].
Washington, DC: U.S. Department of Educa-
tion. Retrieved May 7, 2004, from http://
www.ed.gov/pubs/ReducingClass/title.html
Raimondo, H., Esposito, L., & Gershenberg, I.
(1990). Introductory class size and student per-
formance in intermediate theory courses. Jour-
nal of Economic Education, 21(4), 369-381.
Roblyer, M., & Wiencke, W. (2003). Design and
use of a rubric to assess and encourage interac-
tive qualities in distance courses. The American
Journal of Distance Education, 17(2), 77-98.
Roblyer, M., & Wiencke, W. (2004). Exploring the
interaction equation: Validating a rubric to
assess and encourage interaction in distance
courses. Journal of Asynchronous Learning Net-
works, 8(4). Retrieved December 21, 2004, from
http://www.sloan-c.org/publications/jaln/v8n4/
v8n4_roblyer.asp
Schonlau, M., Fricker, R., & Elliot, M. (2001). Con-
ducting research surveys via e-mail and the
Web. Retrieved December 8, 2004, from http://
www.rand.org/ publications/MR/MR1480
Simonson, M. (2004). Class size: Where is the
research? Distance Learning, 1(4), 56.
Sellani, R., & Harrington, W. (2002). Addressing
administrator/faculty conflict in an academic
online environment [Electronic version]. The
Internet and Higher Education, 5(2), 131-145.
Retrieved December 8, 2004, from http://www
.sciencedirect.com/science/journal/10967516
Simonson, M., Smaldino, S., Albright, M., &
Zvacek. S. (2003). Teaching and learning at a
distance: Foundations of distance education
(2nd ed.). Upper Saddle River, NJ: Prentice-
Hall.
Sorensen, C., & Baylen, D. (2000). Perception ver-
sus reality: Views of interaction in distance edu-
cation. Quarterly Review of Distance Education,
1(1), 45-58.
Southern Association of Colleges and Schools.
(2000). Best practices for electronically offered
degree and certificate programs. Retrieved
April 27, 2004, from http:// www.sacscoc.org/
pdf/commadap.pdf
Sugrue, B., Rietz, T., & Hasen, S. (1999). Distance
learning: Relationships among class size,
instructor location, student perceptions, and per-
formance. Performance Improvement Quarterly,
12(3), 44-57.
Thomas, E. (1984). Class size. Eugene, OR: ERIC
Clearinghouse on Educational Management.
(ERIC Document Reproduction Service No.
ED259454)
Toth, L., & Montagna, L. (2002). Class size and
achievement in higher education: A summary of
current research. College Student Journal, 36(2),
253-260.
United Nations Educational, Scientific and Cultural
Organization. (1997). International standard
classification of education: ISCED 1997.
Retrieved February 2, 2005, from http://
www.unesco.org/education/information/
nfsunesco/doc/isced_1997.htm
University of Illinois. (1999). Teaching at an inter-
net distance: The pedagogy of online teaching
and learning. Retrieved June 3, 2004, from http:/
/www.vpaa.uillinois .edu/reports_retreats
/tid_final-12-5.pdf
Visser, J. (2000). Faculty work in developing and
teaching Web-based distance courses: A case
study of time and effort. The American Journal
of Distance Education, 14(3), 21-32.
Vrasidas, C., & McIsaac, M. S. (1999). Factors
influencing interaction in an online course. The
American Journal of Distance Education, 13(3),
22-36.
Wagner, E. (1994). In support of a functional defini-
tion of interaction. The American Journal of Dis-
tance Education, 8(2), 6-29.
Western Cooperative for Educational Telecommu-
nications. (2000). Statement of the Regional
Accrediting Commissions on the evaluation of
electronically offered degree and certificate pro-
grams and guidelines for the evaluation of elec-
tronically degree and certificate programs.
Retrieved January 14, 2004, from http://
www.wcet.info/resources/publications/
Guidelines.PDF
Wirt, J., Choy, S., Rooney, P., Provasnik, S., Sen,
A., & Tobin, R. (2004). The condition of educa-
tion 2004 (NCES 2004–077). Washington, DC:
National Center for Education Statistics.
Retrieved March 3, 2005, from http://
nces.ed.gov/pubs2004/ 2004077.pdf
Yacci, M. (2000). Interactivity demystified: A
structural definition for distance education and
intelligent computer-based instruction. Educa-
tional Technology, 40(4), 5-16.