Paper Accepted for Publication in Selected Papers from the 12th Intl. Conf. on College Teaching and Learning (Jack
A. Chambers, Editor). Jacksonville, FL, April 17-21, 2001: Center for the Advancement of Teaching and Learning.
Virtual Worlds in Large Enrollment Science Classes
Significantly Improve Authentic Learning
Phillip McClean, Bernhardt Saini-Eidukat, Donald Schwert, Brian Slator, Alan White
North Dakota State University
An emerging pedagogy, authentic instruction (Brown et al., 1989), defines an approach
that treats content as support knowledge needed to solve challenging, real-world problems.
Authentic instruction places the learner in the environment of the doer, and while in that
environment the learner assimilates the practices and beliefs associated with a particular
discipline. This assimilation is a direct result of the learner performing activities unique to that
discipline. For example, to learn cell biology, the learner would visit a lab filled with the
equipment and reagents and perform authentic cell biology experiments.
Unfortunately, to offer such an experience to a large number of students is prohibitively
expensive. The challenge for geology teaching is even more daunting: it is impossible to
transport hundreds of students to the many physical environments necessary to perform those
relatively simple, but unique experiments that are necessary to support problem solving.
Without these economic limitations, authentic learning would support the active engagement of
student-centered collaborative learning of the principles and practices that define science and its
discipline. So what options are available to provide authentic learning in our cost-constrained
Properly designed computer-aided courseware that render cell biology laboratories or
geological worlds can potentially support authentic instruction anywhere a monitor is available.
Importantly the cost of the authentic experience is then pro-rated into the technology budget of
the school. Therefore the educator is challenged to either identify off-the-shelf software that
supports the authentic learning pedagogy or develop new courseware. At North Dakota State
University (NDSU), the World Wide Web Instructional Committee (WWWIC) chose the latter
approach. Here we provide evidence from large experiments that graphical or text-based virtual
worlds designed to support authentic instruction in cell biology and geology can significantly
improve the student’s ability to solve authentic problems associated with those two disciplines.
DESCRIPTION OF THE STUDY
The WWWIC (http://www.ndsu.edu/wwwic) is a multi-disciplinary group of NDSU
faculty engaged in the development of virtual/visual worlds for science education that
communicate both the scientific method and discipline-specific content and are role-based, goal-
oriented, learner-oriented, immersive, and exploratory (Slator et al., 1999). The worlds employ
consistent elements across disciplines and, as a consequence, foster the sharing of development
plans and development tools. Of particular interest for the experiments described here are two
virtual worlds, the Virtual Cell (White et al., 1999) and the Geology Explorer (Schwert et al.,
1999). Each is hosted on the Internet and has the capability of multi-user interactions. The
common technology is a LambdaMoo (MUD Object Oriented; Curtis, 1998) server and database
that contains the contextual material (help file and experimental output data) and controls the
single and multi-user connectivity and interactivity. For the Virtual Cell (http://vcell.ndsu.edu),
LambdaMoo also manages the three-dimensional display of VRML worlds representing a virtual
laboratory, the interior of the cell and its organelles. The display of the world and its direct
interactivity is managed by a Java applet. The version of the Geology Explorer used in the
experiments described here utilized the textual interface features of LambdaMoo. A graphical
version of the Geology Explorer is now available (http://oit.cs.ndsu.nodak.edu/).
The basic experimental design for the Virtual Cell and Geology Explorer experiments
were nearly identical. Student volunteers were recruited from a large-enrollment introductory-
level general science class (General Biology or Physical Geology) with the offer of extra credit
points. The Virtual Cell experiment was performed with two sections taught simultaneously by
different instructors, whereas the Geology Explorer experiment was involved a single section.
Each student volunteer and non-volunteer completed a pre-treatment scenario-based assessment
exercise. These exercises were problem-based questions specific to one of the disciplines. (Visit
http://www.ndsu.edu/wwwic/vc/evaluation/eval2.htm for the Virtual Cell scenarios.)
Within each course, the volunteers were assigned randomly to one of two groups based
on data from a student-completed survey (http://www.ndsu.edu/wwwic/vc/evaluation/eval1.htm)
that characterized their computer literacy, gender, and prior laboratory experience. The Virtual
Cell group completed authentic Organelle Identity and Cellular Respiration activities in the
virtual world, and the WWW group performed two computer-based World Wide Web exercises
that required a similar amount of time with computer-based activities. The Geology Explorer
group was assigned a single authentic mineral identification activity, and the corresponding
Alternative completed an exercise requiring WWW-based activities. For both experiments, non-
volunteers formed the control group and performed no additional educational activities. About
one month after the activities were completed and just prior to the end of the semester, students
completed post-intervention scenario-based assessments. A total of 334 and 368 students
participated in the Virtual Cell and Geology Explorer experiments, respectively. Multiple
student graders trained against a standard approach scored the pre- and post-intervention scenario
assessments. For both experiments, a score of 100 was indicative of the score a professional in
the field would obtain. Because a significant correlation of scores for the Virtual Cell graders
was observed, the mean score was used as the experimental observation. A separate goal of the
Geology Explorer experiment was to evaluate fourteen assessment scenarios (as opposed to just
one used for the Virtual Cell experiment). Because of smaller sample size per scenario, reliable
correlations of grader scores within a scenario could not be obtained. Therefore, the scores of
each grader were evaluated separately.
Two-way analyzes of variance (ANOVA) were performed to determine if any significant
Virtual Cell treatment effects could be detected. The observation that the means between the two
class sections were not significant implies the teaching approaches by the instructors in the two
sections did not confound the student scenario assessments scores. The ANOVA also
demonstrated the post-intervention mean score of the Virtual Cell group was significantly higher
than the corresponding score for the WWW and Control groups (Table 1). This large experiment
Mean post-intervention scenario scores for 1999 Virtual Cell
experiment with NDSU Biology 150 (General Biology) students.
Mean Organelle Identification Cellular Respiration
Treatment population sizes are: Control=145; WWW=94; and Virtual Cell=93.
Within any column, any two means followed by the same letter are not significantly
different at P=0.05 using the LSD mean separation test.
clearly demonstrates the Virtual Cell experience had a significantly positive effect on the ability
of students to solve problems in the mode of a cell biologist. The fact that the Virtual Cell group
mean was significantly higher than the WWW group strongly suggests the improved ability was
not simply the result of computer-based time-on-task, but rather was directly related to the
Virtual Cell experience. In addition, the results demonstrate the WWW group mean scores were
significantly higher than those of the control group. Because of several confounding factors, it is
more difficult to adequately explain these mean performance differences. This difficulty,
though, does not diminish the significant improvement in performance demonstrated by the
students who used the Virtual Cell.
For the Geology Explorer experiment, the data was analyzed by Analysis of Covariance
to adjust for variation associated with pre-intervention scenario assessment score and variation
associated with varying degree of difficulty of the fourteen post-scenario assessments (versus
one for the Virtual Cell experiment). Consistent with the results of the Virtual Cell experiment,
the mean post-intervention scores for those students completing Geology Explorer goals were
significantly higher than the alternative and control groups (Table 2). Although the means of the
three graders varied, the trend of Geology Explorer students performing better on the authentic
assessments was consistent. Unlike the Virtual Cell experiment, the control and alternative
group mean scores were not significantly different.
Mean post-intervention scenario scores for 1998 Geology Explorer
experiment with NDSU Geology 105 (Physical Geology) students.
Mean One Two Three
25.1a 25.5a 44.5a
29.3a 27.0a 42.6a
40.5b 35.4b 53.4b
population sizes are: Control=195; Alternative=95; and
Geology Explorer=78. Within any column, any two means followed by
the same letter are not significantly different at P=0.05 using the
Duncan’s multiple range mean separation test.
CONCLUSIONS AND RECOMMENDATIONS
Given the consistent experimental approach we have employed, we conclude the use of
virtual worlds designed as active, authentic learning environments can positively impact student
learning. Although we describe a computer-based approach to authentic learning, in general we
recommend authentic learning as a component of science learning. The one-time cost of the
technology is expensive, but it is not a recurring cost such as annually funding laboratory
exercises or transporting students to remote learning sites. For upper division courses with
smaller enrollments, these traditional approaches may still be economically feasible, but we
would advocate the transitioning to a virtual world approach for larger enrollment classes.
This research was supported in part by National Science Foundation grants DUE–
9752548, EAR-9809761, and DUE-9981094.
Brown, J. S., A. Collins, A., & Duguid, P. (1989). Situated cognition and the culture of learning.
Educational Researcher, 18, 32-42.
Curtis, P. l. (1998). Not just a game: How LambdaMOO came to exist and what it did to get
back at me. In C. Haynes & J. R. Holmevik (Eds.), High Wired (pp. 25-42). Ann Arbor,MI:
University of Michigan Press.
Schwert, D.P., B.M. Slator, & B. Saini-Eidukat. (1999). A virtual world for earth science
education in secondary and post-secondary environments: the Geology Explorer (pp 519-525).
Proceedings of the International Conference on Mathematics/Science Education &Technology,
March 1-4, San Antonio, TX.
Slator, B.M., P. Juell, P. McClean, B. Saini-Eidukat, D. Schwert, A. White, & C. Hill. (1999).
Virtual worlds for education. Journal of Network and Computer Applications, 22:161-174.
White, A., P. McClean, & B. M. Slator. (1999). The Virtual Cell: an interactive, virtual
environment for cell biology (pp 1444-1445). Proceedings of the World Conference on Ed
Media and Hypermedia (ED-MEDIA 99), June 19-24, Seattle, WA.