Content-type: text/html ~ Stephen's Web ~ Interactivity: Another Tack On It (2)

Stephen Downes

Knowledge, Learning, Community

Oct 25, 1998

In this series:

- Interactivity and Best Practices in Web Based Training
- Interactivity: Another Tack On It (Part 1)
- Interactivity: Another Tack On It (Part 2)
- Interactivity: Another Tack On It (Part 3)
- Interactivity: Another Tack On It (Part 4)

Posted to WWWDEV 26 October 98

Andrew Doherty writes, I totally agree with your comment that "It is useful to quantify degrees of interactivity, because such a measurment will relate also to the effectiveness of the medium as a learning tool". But I don't think it is as simple as equating the degree of interactivity to learning outcome (i.e. high interactivity = more effective learning material).

And I did not and would never claim any such thing. At best, all I could say is that, all other things being equal, more interactivity tends to lead to more effective learning material. Interactivity is only *one* factor in any measure of the effectiveness of a learning activity.

Surely you have read a book (low form of interactivity) and learned quite a lot from it? This, of course, depends on the quality of the book. Similarly, the *quality* of the interaction between web users must be considered. You primarily ignore this in the first part of your email and measure interactivity in terms of bytes of data transferred.

I ignore this in the first part of my email because I wanted to present the concept in quanta which were easily grasped and understood by my readers. As you recognize below, I recognize that the quanta I was using were not appropriate to the task, and a more precise set of quanta needs to be defined.

But then in your last comment you recognise the need to measure not bytes but the *information* (and quality of information) transferred between users: (Strictly speaking, such a measure of interaction should be a measure of the *information* sent each way, as opposed to a measure of the raw number of bytes. I am using byte figures to make the calculations more transparent.) But how do we measure the quality of information transferred between users? I don't think you can if you are talking about measure the interaction between a group of students engaged in a "chat".

If we state that the quantity of information cannot be in any way measured, then we cannot distinguish in any way between high levels and low levels of interactivity. However, since (prima facie at least) we can make such an evaluation, then there must be at some level some means of quantifying information.

This is no place for a treatise on quantifying information (if you want such a treatise, Fred Dretske's "Knowledge and the Flow of Information" is a good place to start).

But, basically, a transfer of data counts as information is it reduces the number of possible states of affairs from the receiver's point of view. The quantity of the information transferred is a function of the degree to which the number of possible states of affairs is reduced.

For example, suppose you knew that Susan was wearing a dress, and that she has in her closet a red dress, a blue dress, and a green dress. The number of possible states of affairs (relative to you, relative to Susan's choice of dresses) is three.

If someone told you, "Susan is wearing the green dress", then the number of possible states of affairs has been reduced from three to one, or 33 percent of the original.

If someone told you, "Susan is not wearing the red dress", then the number of possible states of affairs has been reduced from three to two, or 66 percent of the original.

Both sentences contain information, but clearly, the first sentence contains more information than the second.

Notice that the amount of information conveyed is realitive to the receiver. If I told Susan, "You are wearing a green dress", and Susan already knew she was wearing a green dress, then I have not transferred any information, since the number of possible states of affairs both before and after my statement is one.

Thus also we can measure the interactivity of learning systems: though of course the calculation of quantities of information is by no means as simple as in the examples just given (which is why I used a much simpler quantification of information).

Returning now to evaluating learning materials: the effectiveness of learning materials consists essentially in the amount of information such a system can successfully transfer to the viewer. The qualification, "successfully", is important here. By "successfully" I mean something like "retained by" or "can be used by" or "internalized by" the viewer.

A simple information dump would, in ideal circumstances, result in the most effective learning, because a simple information dump would result in the largest possible reduction of possible states of affairs. However, the circumstances are never ideal. It is generally not possible for an individual to internalize a simple information dump.

Generally, when a viewer interacts with the information being transmitted, that viewer's capacity to internalize the information is increased. Thus, the more interaction there is, the more information may be successfully transmitted. For example, if I simply receive a stream of information, I may be able to retain 10K bits of information. However, if I interact with that information, I may be able to retain 20K. The amount of information an individual may retain varies with age, experience, and level of education.

Thus, a measure of the quality of learning materials would include, first, the quality of the information transmitted, and second, the degree of interactivity afforded by the means of transmission.

By "quality of information", I mean the ratio between data which (in ideal circumstances) reduces the number of possible states of affairs for the viewer, and the data which does not reduce the number of states of affairs. Or to put the same point another way, the ratio between the data which is new to the viewer, and the data which is not new. Or to put it another way, the ratio between signal and noise.

A high degree of interaction is of course insufficient for quality learning materials if the data being transmitted is mostly noise. This is why undirected chat sessions are poor learning tools.

A high degree of information without interaction is also generally insufficient for quality learning materials, because the quantity of information which may be internalized is too low. This is why lectures are often poor learning tools.

Therefore you haven't provided any measure of interactivity here!

Now that's a bit harsh, don't you think?

Stephen Downes Stephen Downes, Casselman, Canada

Copyright 2024
Last Updated: Jul 18, 2024 7:28 p.m.

Canadian Flag Creative Commons License.