Attending Class / Offline Class

What's good about attending class / offline class?

A brief history of educational media

The development of educational media has greatly influenced educational forms. Many realize that projectors and computers popularized slides; recording, broadband, and streaming popularized course recordings / online courses / MOOCs.

It's hard for us now to imagine education before slide projectors or even printing. But technology's transformation of education is sometimes slower than we think. After a new technology is born, it may take considerable time to go from hard-to-use to easy, from expensive to cheap.

For example, an educational-history paper discussed Gutenberg's printing press and its effect on early European universities. It argued that "Printing greatly expanded the provision of information but did not change the way people learn." That is, at least when printing first became common, educational forms didn't fundamentally change; "memorization" persisted for decades or even a century after printing improved and books proliferated.

In 1913, Edison declared that within ten years films would replace books in academia and upend the school system. A century later, that has not happened.

Professor Li Xiaoming's 2004 article "What does the democratization of network video production bring to education?" also felt video might transform education — prophetic. But whether live audio/video or uploaded recordings, around 2000 costs were extremely high; the MOOC boom came after 2015.

References:

Reiser, R. A. (2001). A History of Instructional Design and Technology: Part I: A History of Instructional Media. Educational Technology Research and Development, 49(1), 53-64.
Li Xiaoming. What does the democratization of network video production bring to education? Computer Education, 2004.
Gavin Moodie (2014) Gutenberg's effects on universities, History of Education, 43:4, 450-467, DOI: 10.1080/0046760X.2014.930186

Compared to improving docs and providing tutoring, what benefits does a TA "recitation" bring to students?

Are substitutes for classroom lectures necessarily good?

At first glance, good self-study materials that substitute for classroom lectures seem easy to find — e.g. the CS self-study guide index of excellent overseas courses or classic textbooks. So what's the point of attending class?

Actually, these "substitutes" have drawbacks:

Textbooks: textbooks/notes/docs are "teaching materials" here. The biggest issue may be lacking good self-study material. Domestic textbooks are often too terse; overseas textbook translations vary in quality; undergraduates reading English is tiring. Following the traditional publishing cycle, textbooks usually don't include the latest content; teachers must supplement. (Possible measures: accumulate docs/notes over time? Some excellent textbooks emerged from years of notes. If you use online, updated docs/notes and maintain them, you can keep content current — some assignment docs do this well. But it costs teacher/TA effort.)

Slides: originally to aid teaching, not necessarily suitable for self-study. (Possible measures: make slides suitable for self-study/review, even make "class version" and "self-study version"? To facilitate discussion, put slides in an online doc where students can annotate to ask questions?)

Online courses: may not fit students' background, may not fit the course's exam scope. (Possible measures: recommend good online courses? Even down to each class, recommend videos for relevant knowledge points?)

Local teachers can tailor for local students: prepare before class for their background and interests, adjust pace during class, iterate after class based on feedback. MIT's recordings are prepared for MIT students' prerequisites and training goals, aiming to maximize MIT students' education. At our school, students' foundations, ability, interest, and difficulties differ. If teachers prepare accordingly before class and adjust during, students feel it's better than the best other school's online course. Of course, this is harder than it sounds — you need teaching fundamentals and to make good use of student feedback.

Failure to Disrupt: why technology alone can't transform education.

Discussion topic 1: How to collect and use feedback on "classroom teaching"? What are the pros/cons of each measure?

Possible answers: teachers can collect feedback via many means. Pre-semester surveys, distributing/collecting preview worksheets to learn students' perceived difficult points, watching students' eyes in class, chatting at breaks, judging absorption via assignments, using Rain-Classroom quizzes. Combine with teacher/student preferences. Also, have TAs audit the class, or invite other teachers to audit for feedback.

Theory Discussion: Cognitive Load

Cognitive load is the use of the brain's "working memory" during a cognitive process. Analogize the brain to a computer; psychology's "working memory" is this computer's cache. In plain terms, if a cognitive process requires "multiple turns of thought" and handling many things in working memory simultaneously, the load is high. Learning is a cognitive process.

So reducing unnecessary cognitive load in class yields better teaching. Based on simple assumptions, researchers summarized effective load-reduction measures.

"Information store principle": experts in tasks like programming often solve problems in the brain via pattern matching. They remember more "patterns" than novices, so lower load. This may underlie "practice makes perfect" and "problem drilling."
"Borrowing and reorganizing": most of our long-term memory information comes from others' long-term memory; we recombine existing knowledge with others' information. This explains why collaborative learning works.
"worked example effect": step-by-step demonstration of solving a problem is a low-load way for students to learn to solve it.
"split-attention effect": avoid requiring students to attend to multiple places at once. E.g. if a slide has both text and a figure, combining them is lower load than separating.
"expertise reversal effect": as students' knowledge rises, the most effective level of guidance changes; the less prior knowledge, the more detailed guidance needed.

Reference: John Sweller. 2016. Cognitive Load Theory and Computer Science Education. SIGCSE '16 — an abstract of the founder's SIGCSE 2016 keynote.

A note on John Sweller's 2016 keynote.

https://www.youtube.com/watch?v=gOLPfi9Ls-w — video of Sweller on cognitive load theory.

Offline class vs. online class vs. watching recordings

The full online teaching during the COVID spring 2020 had far-reaching effects. Whether online / blended-hybrid / offline is better is unsettled; it relates to teacher/student preferences.

For example, a national survey of college students listed some possible defects of pandemic online teaching, showing large variation across courses/teachers.

Students rely too much on playback, thinking they can re-learn, so class efficiency drops (~37% agree, 21% disagree).
Online communication is less direct than offline (~35% agree, 23% disagree).
Lack of teacher's on-site guidance/supervision, lax discipline (~43% agree, 17% disagree).
Teacher can't immediately answer students' concerns (~43% agree, 16% disagree).
Teacher can't promptly know students' mastery (~49% agree, 13% disagree).
Teacher can't immediately know students' learning state (~49% agree, 13% disagree).

Reference: Wu Daguang, Li Wen. Stage-based features of large-scale online teaching in Chinese universities — based on questionnaire research of students, teachers, and academic staff. East China Normal University Journal (Education Sciences), 2020.

Another study based on Tsinghua evaluation data: in online teaching, students' score for "clarity of teaching goals" dropped vs. offline; lower-grade students more want in-class, juniors adapted better. Also, good teachers got good evaluations both online and offline — so the determinant isn't online/offline but instructional design and teacher investment.

Reference: online vs. offline undergraduate teaching quality comparison — based on Tsinghua teaching evaluation data. China Educational Technology, 2022.

If a course offers online attendance/recordings, should we still encourage offline attendance? Depends on the teacher. If the teacher thinks online/playback is also good learning, provide those channels and don't worry about offline attendance. If the teacher thinks offline is the best / only enters a good teaching state with students present, consider not providing online/playback (for special cases who truly can't attend offline, provide individually). Some students prefer the offline atmosphere; some find speeding through recordings more efficient. Sometimes the school requires offline attendance or providing online channels.

From the teacher's perspective, reasons to want students offline:

Good courses need polishing. Teachers polished under past technical conditions and are used to old modes; migrating to new tech costs more, not necessarily better. E.g. Gilbert Strang's last lecture is the most traditional blackboard teaching. (Recording tech lets the world see it.)
If the classroom is empty/nearly, teachers feel frustrated, lose face-to-face feedback, harming teaching quality and enthusiasm.
Writing slides in exhaustive detail, or detailed notes, is time-consuming. Many teachers think PPT aids a talk, not a script/self-study material; academic talks don't put everything on slides. See Armando Fox: making good slides.
(To be added.)

But students won't attend offline just because "the teacher wants me to." Sign-in, in-class quizzes put bodies in the room, not minds.

Classroom interaction / active learning

Ways to increase offline appeal/effectiveness include adding questions/quizzes to traditional lecturing, and reducing lecturing in favor of more active learning.

After adding activities, students participate offline more and understand some points more deeply, but the efficiency of knowledge transmission may drop.

Research on teacher-student interaction improving classroom teaching: good interaction improves effectiveness (Zhang Jinghua. Teacher-student interaction in undergraduate classroom teaching: current situation and effects. Education Development Research, 2019).

Since the goal is better learning, collect student feedback on interaction forms — direct evaluation, and comparison of learning effects, e.g. use a new form for point A, traditional lecturing for point B; then via quizzes/assignments compare mastery of A and B, partly revealing the form's effect.

Discussion topic 2: Based on your experience as student/teacher, which three classroom-interaction forms in an active-learning list are most suitable/effective/popular? Should interaction (e.g. Rain-Classroom answers) be graded?

Reference: UW Active Learning activity list.pdf

Recitation

Besides going over assignments, recitations can review classroom knowledge.

(If time allows, primarily for TAs, run an extra recitation trial + peer-review activity. This needs a rubric for evaluating recitation quality.)

What makes a good recitation?

Basic quantitative courses widely use recitations. Overseas CS courses often use recitations too. Recitations usually don't introduce new knowledge; they review the week's lecture. Since lecturing pace can't suit everyone, some students can't digest in class; a TA-led review of the week's content and some worked examples helps those who fall behind.

Some add diverse extension content and student choice, e.g. three recitations a week: this-week review, no-exam extension content, and an industry guest talk: Melinda McDaniel. 2023. A la Carte Recitation: Allowing Students to Choose their CS Laboratory Experience. SIGCSE 2023.

How to ensure participation? Forced participation brings management issues; not forcing may yield low attendance.

A middle ground: require at least 2-3 offline attendances in the semester. Provide recordings for those with conflicts.

Also, large-lecture recitations or small-group seminars? The latter works better but needs many TAs, which most courses lack.

Discussion topic 3: How to collect feedback on recitations to improve? How to build a rubric for recitation quality?

Reference: