Slashdot Mirror
The Problems With Online Math Classes
dcollins writes "As a college instructor specializing in statistics, I felt compelled to survey one of the massive-enrollment online education courses that are all the rage these days. This summer, it seemed a perfect opportunity when Udacity unveiled Introduction to Statistics by founder Sebastian Thrun (of Google autonomous car fame). Having taken the entire course through to the final exam, my overall assessment is: It's amazingly, shockingly awful. Some nights I got seriously depressed at the notion that this might be standard fare for college lectures encountered by many students during their academic careers. I've tried to pick out the Top 10 problems with the course structure and address them in detail."
Some teachers who have tried using Khan Academy in the classroom (I read about it offline so no link, sorry) said they thought the best advantage wasn't that the presented material was better but that it freed them up from presenting the information so they could give more personal assistance.
Imagine that in maths class rather than the teacher standing at the board for 30 mins explaining an example they could let a small group watch the video while helping another small group who just finished watching it. Alternatively the time freed up could be used to customise homework to stretch those at the top and bring up the quality of those at the bottom.
Finally add additional functionality on and allow access to the vids at any-time and anyone struggling with 'integration' could look back over that video description, perhaps access an online/skype-esque tutor service and/or post what they are stuck on for the teacher who could arrange very brief one-to-one sessions to address specific issues.
I think way too much focus on these courses is on how they 'replace' teachers. Sure there is some scope for that but I think we can get far more benefit by augmenting classroom teaching.
I took Prof. Thrun's & Prof. Norvig's course, "Introduction to Artificial Intelligence" when it was first offered. I'm pretty big on self-study, and I rely on instructors to provide efficient direction (a syllabus, specific reading material), a mechanism for self-evaluation (exercises, means to validate results, etc), and finally, a source of answers when I have questions. In a perfect world, online courses seem to be a good fit for my personal needs, so I dove in with relish.
However, I found some of the same general problems the blog post referenced;
- the content (speech, writing) was often sloppy and confusing, it did feel unplanned.
- concepts that were introduced were not explained in their entirety.
- the vocabulary used to describe a new idea was fairly mutable, or inconsistent.
- there were often instances requiring sizable leaps of intuition combined with formal mathematical knowledge to complete exercises which had previously only been provided in a "fill in the numbers" format in previous examples.
In addition, I found no clear mechanisms for self-evaluation. We had to wait a week just to see the results of previous tests, etc. I also thought the quiz interface was childish and poorly done, but that's mostly just a look and feel issue.
I also took Prof. Ng's "Machine Learning," class at the same time. In contrast, I found that Prof. Ng provided:
- Writing was clear, dialog was polished, vocabulary was explicit.
- Concepts were introduced, explained (in both a practical and intuition-focused form), demonstrated and expanded upon.
- Exercises were given to students in the form of example data, algorithms to implement, and with additional suggestions on how to 'play' with them to produce different results and gain an intuitive grasp of the information. Unlimited resubmission of exercises with an automated grading system made evaluation of different mechanisms simple.
- Quizes were more polished.
I felt like I got a lot out of his class, well more than the AI class.
I feel that the difference between the two was pretty obvious. Prof. Thrun was teaching as if he had a live audience in front of him, and did not modify his instruction style for the lack of interactivity. On the other hand, Prof. Ng taught in a way that minimized the deficiencies of video learning, while leveraging the benefits of online, automated instruction.
In conclusion, I don't think the AngryMath blogger is correct in the assumption that live, in-person instruction is needed. In fact, I'd say the opposite was shown: the closer you get to the style of live instruction, the worse it seems to be to me, and more so when it's online. Of course, I have specific needs from education, and others may prefer different styles.