The current Am Journal of Physics has a very good PER article where they
looked at textbooks vs multimedia. They pointed out that there was no
evidence that any specific feature of texts improves effectiveness. Indeed,
from what I have seen texts run contrary to what we know about learning and
as such are probably not very effective.
But there is a lot of research on multimedia so they created some multimedia
lessons and compared them to equivalent text. They also reduced the
multimedia to scripts and used that as a third comparison. The multimedia
was designed according to known research. Students volunteered for the
study, and were paid to do the lessons. They were randomly assigned to on
of the 3 types of lessons, and of course the lessons were done under
supervision to assure that the students actually completed them.
The result was about 15% higher learning for the multimedia, or about 0.75
effect size. This translates into approximately 0.3 Hake gain compared to
the standard text lesson. The script lesson on paper from the multimedia
was intermediate in results.
Students expressed more likelihood of using the multimedia "text" in
preference to the standard printed text. Since students seldom crack open
the textbook except to do assigned problems this alone would be an
improvement. But I think the more interesting part of this is that the
scripted text based on multimedia did better than the standard text. This
points to the fact that standard texts are not written according to the way
that people actually think and learn.
I have been fairly skeptical about multimedia because it has usually been
created in a seat of the pants fashion without any research guided design.
People have been claiming that it works better but most research shows it
does not. There are few papers that show signifantly better results, and
the ones that I have seen restrict students to research designed media. So
just throwing multimedia at students will not work. But it is possible to
design multimedia that actually works.
This article is also an example of how education research can help us make
teaching a science rather than a seat of the pants endeavor.