VOL. 14, No. 2, 85-96
This article reports the delivery of a completely online version of an introductory statistics course. STAT 101 has been offered at Simon Fraser University in seven successive trimesters during September 1997-December 1999. The course has minimal mathematics prerequisites and yet is a serious introduction to the concepts of statistics. Verbalization, visualization, conceptual understanding, and problem-solving are emphasized, with some efficiencies gained by relying on computer software for graphs and calculations. In the descriptive part of the course, topics include bivariate data, time series, categorical data, and data presentation; in the inferential part, sampling, study design, and inference (including the simplest ANOVA and regression) are covered.
The motivation for offering an online version of the course is to improve on the educational quality of the correspondence version of the course in a similar cost envelope. To achieve this the use of online two-way communication over the Internet has been added to the original features of the correspondence course—the latter used self-study, assignments, and telephone and postal communication as the only pedagogic tools. Unlike many other initiatives in online instruction, the Internet has not been used in this course for the supply of data or information. The use of the Internet as a communication tool substituting for face-to-face contact has been exploited.
The main obstacles to overcome in the design of an online course are: compatible and adequate software and hardware for communication; lack of two-way communication with others associated with the course; avoidance of viruses; and identification of the person submitting the material. The design described here has overcome these impediments as I show below.
At the beginning of the course students are allocated to study groups of four or five students, and these groups usually do not change. In a study group students can communicate with each other without their online conversation being accessible by other study groups. These study groups are responsible for answering fairly open questions from each week’s material (e.g., “Which technique is best for graphing a univariate data distribution?). The study group appoints a moderator, who is responsible for submitting the group consensus, and this moderator job rotates through the study group. These group assignments count for 15% of the course grade. Participation in these study group assignments and in other course conferences counts for another 10%. These percentages have been high enough to coerce a high degree of compliance in study group activity; for example, a recent class averaged a grade of over 80% on this portion of the course grade.
There are also assignments submitted by individuals, which count for another 15% of the grade. Students submit assignments electronically, and responses are returned electronically. This submission process is handled using a conferencing system called FirstClass® published by SoftArc (1998). Each message has the capability of including graphs. Attachments were not used, in order to avoid virus problems. The problem of communicating equations is not a big one in this course. Most communication is a record of verbalizations or graphs, which FirstClass handles easily.
The FirstClass system allows the students access to several online conferences connected with the course. Here is a list of the ones I use in this course:
The dialogue in these conferences ranges from the informal back-and-forth chat (1 and 2 above) to the more formal one-way presentations (5 and 6 above). The Study Group conference has a formal goal—the submission of the answers to group assignments—but the process is quite informal. Feedback on assignments is sent to students individually via e-mail, with their electronic submission annotated with comments from the instructor. The more informal conferences allow students’ personalities to emerge, reassuring students that they are being educated by people rather than machines. This is a huge improvement on the correspondence course, where the only communication was via handwritten note at intervals of a few weeks.
A mid-term test and final exam are given in person, and these components count for 60% of the course grade. Students are required to pass these in-person tests. Remote writers arrange invigilation by a qualified individual, usually a high school teacher.
The FirstClass system manages the study conferences and the questions and answers between students and the instructor. Most questions and answers are accessible by the whole class, just as they would be in a classroom. Private conversations are possible by ordinary e-mail, but students are told they will not receive participation credit for this, to encourage them to use the shared conferences for questions.
There is a read-only conference in which the instructor can alert the entire class to certain important ideas, administrative changes, old examinations, or assignment solutions. A useful feature of FirstClass is that for any particular posted message the instructor can easily get a list of who has accessed the message and when. It is thus possible to tell who will have missed an important message.
FirstClass is available for Macintosh or Windows PCs. Students download the software at no cost, and the institution pays a one-time fee per student account. The accounts are reusable in subsequent semesters. The course text is Griffiths, Stirling, and Weldon (1998).
The main input for estimating the cost of the course is the online time of the instructor. The instructor submits bi-weekly hours in order to receive remuneration. This includes time for marking including response to students, online questions and answers, preparation of notices, e-mails for private interaction, and evaluation of participation by students. The total hours charged at the end of the course are divided by the number of students completing the course, to obtain the hours per student. Over the seven trimesters 97-3 to 99-3, the numbers of completing students were 41, 21, 23, 30, 37, 39, 41, and the per-student hours spent by the instructor were 13.8, 14.3, 11.3, 9.1, 8.2, 8.0, and 5.2 in chronological order. The reduction in instructor time was due to the elimination of unforeseen problems due to this new mode of instruction, the switch to FirstClass for the most recent four semesters, and the provision of more efficient management guidelines to the instructor. Completion rates for the course were approximately 70% initially, but have increased to over 80% in recent semesters.
Knowing the hours per student in the online course allows a rough cost comparison with the lecture version. Using rough figures, I computed the operating costs in both modes as a function of the enrollment. Assuming that the variable cost per student (mostly marking and tutorial assistance) is about twice what it is for students in a lecture course, but that the cost of a lecturer is about six times the cost of providing the supervision of the online course, it works out that for courses with enrollment of about 70 or fewer, the costs of the online course are less than the costs of the lecture version. With a huge class, if all are listening to a single lecturer, the lecture version is clearly less expensive. Of course, the infrastructure costs of both modes are not included in this comparison and may be important for a comprehensive analysis.
The idea of using the Internet to add socialization to an otherwise lonely correspondence course seems to have worked well: students seem to enjoy the easily arranged nonsimultaneous online communication. Achievement on the examinations is comparable to the lecture versions of the course. When students have a choice of the online course and the conventional lecture version, only about 20% opt for the online mode. But for those with scheduling problems, or who live far from the university, it does fill a need. Moreover, from the perspective of the university, the course model described here is less expensive to deliver than the lecture version, once the start-up problems are solved, for courses of fewer than about 70 students.
SoftArc Inc. (1998). FirstClass Version 5.506. Detailed information available: info@softarc.com
Griffiths, Stirling, & Weldon. (1998). Understanding data: Principles and practices of statistics. Brisbane: Wiley.
K. Larry Weldon is an associate professor of statistics at Simon Fraser University, Burnaby, BC, Canada. He received his doctorate in statistics from Stanford University in 1969. weldon@sfu.ca