A colleague shared the following priceless reflections, a letter to Chris Edley from July 28:
Dear Dean Edley:
I’ve been following with interest what you’re saying in the press about UC online education.
I teach Statistics N21, the first online course at Berkeley to be approved by COCI [the UCB faculty Committee on Courses of Instruction]. It was approved in 2007. I’ve been teaching it for four years, this year to 400 students. The current syllabus is here:http://statistics.berkeley.edu/~stark/Teach/S21/Su10
Statistics N21 a gateway course: probably one of the first 10 you would want in your pilot. It satisfies major requirements for several departments, and is a “hurdle” course for intended Business majors.
The online course comprises an interactive textbook (SticiGui: http://statistics.berkeley.edu/~stark/SticiGui) that has Java applets to illustrate key concepts, examples and exercises that change when the page is reloaded so that students can get unlimited practice with the material, machine-graded assignments scored using a mastery model, videorecorded online lectures, online and in-person office hours, a discussion board, etc. Every student gets a different version of the online assignments. The final is administered in person. Most students take the final on campus, but about 85 will take off-campus proctored finals this summer, in several countries.
SticiGui has been used at other colleges and universities to teach statistics classes and to teach methodology classes in economics (at CUNY) and political science (at Bard).
But it also has interactive chapters and machine-graded assignments suitable for general education classes: Reasoning and Fallacies, Categorical Logic, Propositional Logic, and Set Theory. It has been used to teach linguistics and logic classes at UCSC and SJSU.
The infrastructure, applets, and so on that I have built could be adapted most easily to teach introductory courses in mathematics, economics, demography, sociology, and similar fields. But I think it would take a considerable amount of work–years of careful attention from devoted faculty–to develop pedagogically sound, interactive content worthy of UC. Even to build a more advanced statistics class using the same plumbing would take a solid year of full-time work.
It has taken about 8,000 hours of my time over 13 years to develop (what I consider to be) pedagogically effective interactive content and assignments. The materials wouldn’t have worked well as an online-only course for at least the first 5 years of development. I used it to teach hybrid classes while I was developing it, starting in 1997. Work continues: I’m building a searchable database of lecture “clips” on individual topics, edited from my webcast lectures. The clips will also be linked to the text where the topics are introduced, and to the glossary.
In a large-enrollment course like Statistics N21, ensuring that students have up-to-date browsers before the class starts and providing technical support during the first week or two of class are virtually a full-time job. (Those are jobs that GSIs and technical staff can help with.)
The “bandwidth” of online instruction is lower than face-to-face instruction: it takes longer to convey the same information, both from instructor to student and from student to instructor. One side effect is that online office hours are less efficient than in-person office hours, so more office hours need to be offered. Online courses therefore need correspondingly more staff, even before factoring in technical support. To hold online office hours at times that are convenient for students in, say, Taiwan, requires working odd hours. For reference, here is the office hour schedule for N21 this summer: http://statistics.berkeley.edu/~stark/Teach/S21/Su10/index.htm#officeHours
I’d be happy to talk to you about what was involved in developing Statistics N21, the resources required to teach it, and what would be needed to do something similar in other disciplines.
Philip B. Stark | Professor of Statistics | University of California
Berkeley, CA 94720-3860 | | statistics.berkeley.edu/~stark