Here is a paper from Marsha Lovett, Oded Meyer, and Candace Thille:
Abstract
The Open Learning Initiative (OLI) is an open educational resources project at Carnegie Mellon University that began in 2002 with a grant from The William and Flora Hewlett Foundation. OLI creates web-based courses that are designed so that students can learn effectively without an instructor. In addition, the courses are often used by instructors to support and complement face-to-face classroom instruction. Our evaluation efforts have investigated OLI courses’ effectiveness in both of these instructional modes – stand-alone and hybrid.
This report documents several learning effectiveness studies that were focused on the OLI-Statistics course and conducted during Fall 2005, Spring 2006, and Spring 2007. During the Fall 2005 and Spring 2006 studies, we collected empirical data about the instructional effectiveness of the OLI-Statistics course in stand-alone mode, as compared to traditional instruction. In both of these studies, in-class exam scores showed no significant difference between students in the stand-alone OLI-Statistics course and students in the traditional instructor-led course. In contrast, during the Spring 2007 study, we explored an accelerated learning hypothesis, namely, that learners using the OLI course in hybrid mode will learn the same amount of material in a significantly shorter period of time with equal learning gains, as compared to students in traditional instruction. In this study, results showed that OLI-Statistics students learned a full semester’s worth of material in half as much time and performed as well or better than students learning from traditional instruction over a full semester.
Of course not every university has students as good as those at Carnegie Mellon. A longer and more general article is here, hat tip for both goes to AT.
Lectures are worthless, who would have thought? Oh wait, everyone figures that out by senior year.
Also, there’s no hyphen in Carnegie Mellon.
LOL! I didn’t have time to fit statistics into my undergraduate studies since I wanted to double major in chemistry and political science (don’t ask why). During the summer between my junior and senior year, I picked up the Schaum’s College Outline on Statistics and methodically worked through it. It taught me everything I needed to know and for a pretty good price (I think it was $4.95 or something like that back in 1969). Of course these days, you can do almost everything within Excel, a far cry from programming old PDP-11 lab computers.
I have to say that contradicts my experience (from many years ago) with taped lectures. Looking at my peers, it really felt that unless there was a human making the effort to teach, the majority of the students has difficulty in making the (considerable) effort to learn.
Same reason why for most students, a good library and a course outline isn’t sufficient to produce learning.
Would be interesting to see this applied to a more diverse set of students to see if the results still hold.
Properly designed online learning courses aren’t like lecture on tape. They provide instantaneous feedback and help a student focus on their individual weak spots rather than letting them fall behind in areas where they are lost or needlessly focusing on material they already understand. It’s more like having a tutor than purely self taught, though it still requires motivation on the student’s part.
Tom West: “it really felt that unless there was a human making the effort to teach, the majority of the students has difficulty in making the (considerable) effort to learn..” In Infovore Tyler says pretty much the same thing. So Tyler should rethink his observations in the book ?
Always fun when an author of a paper shares a name with someone you know who is the farthest thing from an academic (a ski bum, in this case)
Good piece, thanks for sharing.
That surprises me becuase judging from the Florida Virtual Schools class that me son had to take, the state of the art in online education leaves a huge amount of room for improvement. Perhaps Carnegie Mellon’s software is much better than FSV’s.
Could be the software, but I saw from experience that a well-thought-out and well-constructed course (learn this in this way, do this, learn that in that way, do that, read this for edification, etc.) will have better results than a linearly-arranged course. The instructors in these courses were not especially well-prepared with picture-perfect slides or spot-on assignments. They had a few ideas they wanted to disseminate every class and most of the material (didactic or discussion) was focused around those concepts. In some of the boring statistics classes, I saw that there was a tendency to try to teach as many of the statistical concepts as they could. This approach lost a lot of students. On the other hand, when the instructor simply wanted to get across the idea of regression and allowed some discussion around that topic, it seemed a lot more effective.
There’s a lot of snake oil being sold in the online learning sector (remember when distance learning was going to make bricks-and-mortar schools obsolete, because students could learn by watching lectures on TV? Or for that matter, when writing made schools obsolete — why bother taking a class when you can learn the material by reading a book?). Online learning is not going to make classroom education disappear.
That being said, like the book, online learning is having a big impact on how people teach and learn. CMU’s OLI is one of the non-snake-oil projects; the stuff they’re doing is pointing the way to the future of how online learning will affect the classroom. Hybrid education, flipped classrooms, etc. And pure online classes when appropriate (motivated students with specific limited learning goals — online traffic school e.g.).
Help Wanted Ad From the Future:
Looking for a Ph.D tutor from India to help me with my online statistics course.
Willing to pay up to $3/hour.
Bill, here in India there are people on the lookout for an advertisement like this one : ‘Wanted: someone to supervise and write my PhD thesis online. Will pay handsomely, especially if you can also fix the examiners for the thesis”. These guys … I have come across only men in this business … will deliver a thesis online for about $1,500. A couple of hundred dollars they find a “suitable” examiner for your thesis too.
Sign me up.
The PBS News Hour has a couple of reports on teaching college courses to kids at risk of dropping out or performing below grade level. In a border region school, local college instructors go to the high school to teach, and high school teachers are being certified to provide college credit, with all students encouraged to get college credits to meet high school graduation requirements.
It seems to me online courses offer the opportunity for:
- competition among students to achieve
- make a high school teacher able to offer the best instruction in multiple courses
- provide more flexibility to students who often need to help their family struggle
- frees high school teacher to help students deal with non-academic issues
- puts the initial college semester course expectation shock behind them so the college shock is only social
The problem is this doesn’t help the students who can’t see any future through college because they are people who want to make or build tangible things. Lots of workers are needed in telecom which requires connecting copper and fiber and setting up all sorts of black boxes that require special secret codes like IPv6 address prefixes and so on, or how to turn a rod of titanium into $100 screws, a block of steel into an auto part, multiple titanium into a bicycle frame, how to draw blood and then prep it for the blood analyzers, or culture for drug resistant illness. These skills are required to do a job that involves mostly less skilled activity – grabbing boxes of wire, laying out the wire, pulling the wire from point-to-point, etc. The problem is these course are very expensive compared to a course on economics or statistics.
If the cost of teaching “thinking” courses can be cut by 90%, then I hope the spending on “doing” courses can be increased by 400%.
Is there a reason http://oli.cmu.edu/wp-oli/wp-content/uploads/2012/05/Lovett_2008_Statistics_Accelerated_Learning_Study.pdf is not linked?
This really isn’t so different from teaching yourself a course from textbooks, something that I did for most of my higher level high school and lower level undergraduate mathematics courses and in accounting. The main downside to that kind of instruction pedagogically, is that if you get stuck at some point, you can get really stuck while just a little guidance would allow you to get over the hump. If a “hybrid” method overcomes this flaw, then its all good.
Note also that a common mode of online instruction is “internet delivered” but is basically a traditional format. I’ve taught many classes in that mode.
Selection bias…selection bias…selection bias…?
How so? The CMU paper recorded drop out rates, of course; the ratio was pretty much identical. (Something like 1/22 for online and 2/43 for the controls.)
This interested me because 1) I now do some instruction and build course materials, 2) I took Physics and Calculus Self-paced at CMU in the late 70s before the PC era. I’ve also taken and taught blended approaches.
I went through the introduction and first module of Probability and Statistics and was impressed with the design and execution.
The course is divided into 14 modules. The course began with a tutorial on the importance of planning the work for learning then described nonrecognition and the conditions necessary for learning and mastering the material;
1) Assess the task, 2)Evaluate your strengths and weaknesses, 3)Plan the approach, 4)apply strategies and monitor your performance, 5)reflect and adjust as needed.
It contains a syllabus and web based outline. The module began with defined learning objectives. The module is further divided, textbook like, into short focused sections including formative questions. Additional questions are available if one has difficulty with the material. Some of the questions require using Minitab, R, or Excel (student’s choice), and the data along with instructions are provided. It ends with a practical assumptive exercise. The questions may be submitted to the instructor or taken as an open uninstructed course.
The material appears to be about what I would expect from an introductory university Statistics course, though easier reading than a typical college text. I required about 2 hours to work through the module, but although I never took a formal Stats course I am familiar with the material (I also worked examples in both Minitab and R) . I estimate it would have taken me about 6 hours to complete unfamilar material of similar difficulty.
When I took self paced Physics and Calc, the self paced material paralleled the traditional course. It was expected but not required that one would attend the lectures. I never did (and missed out on Hugh Young’s excellent lectures) but my pace was generally about twice the lecture based class. Skipping lectures was worth at least 3 hours per week, the recitations were probably a wash with the physically attended summative exams at the end of each module.
I don’t expect this approach to work for everyone, it will require some self regulation to get the full benefit. It seems far more likely to be useful than the uploaded traditional course material available at universities such as MIT.
The material is good enough that I plan to work through the remainder of this course to learn to better explain this subject to my own students and to learn a thing or two about instructional design.
Comments on this entry are closed.