The March 30 issue of Science included a letter from psychologists David Daniel and Daniel Willingham  that provides an excellent overview of what is known about how students use online textbooks, including an account of what we know about the strengths and the weaknesses of switching from print to electronic delivery. I find that what they have to say is in line with my own experiences. For the past year, I have taught our Single Variable Calculus class using MAA’s online textbook Calculus: Modeling and Application, 2nd edition, by David Smith and Lang Moore. This is a direct descendant of their Project CALC materials that I have used and loved. The current edition is only available as an online textbook.
The overwhelming advantage of online publishing is the cost savings. The cost to my students is $25 apiece. This is charged as a lab fee. In exchange, my students get to download the entire textbook onto their own computers, a benefit that is relatively uncommon among online mathematics texts but which my students appreciate because they are not restricted to using their textbook only when they have internet access.
Smith and Moore’s book is written in html, using MathML for the mathematics (which effectively restricts the web browser in which it is read to FireFox). The authors have done a thoughtful job of separating the chapters into sections that each fit fairly comfortably on a single web page. Several of my students have commented on how much that helps with the readability of the text. But my students have found that reading an online textbook does require a period of adjustment.
I have yet to encounter a student who prefers reading web pages instead of printed pages. Daniel and Willingham cite three different studies that confirm that most students prefer traditional print books. Intriguingly, online texts work very well for young readers. In fact, those learning how to read often do better with online books. The difficulties seem to arise when students need to study and learn from the text, a characteristic that is especially true of mathematics books. For reasons that we do not fully understand but which are well documented, careful reading of an electronic text takes longer and is more fatiguing than trying to learn the same material from a printed text. The research also has found that this effect of greater difficulty with electronic texts is independent of the level of familiarity and experience with e-books.
One of the greatest benefits of online textbooks is the ability to embed links to definitions, animations, and software programs. Smith and Moore’s textbook is liberally sprinkled with links to explorations that are available in Maple, Mathcad, and Mathematica. There also are links to WeBWorK, where a library of problems linked to the sections of their book is available.
My own experience is that students frequently ignore the links to explorations unless I specifically assign them. This is in line with the findings reported by Daniel and Willingham: Students often find such ancillary material distracting at best, confusing at worst. Following these links often leads to loosing the thread of the conceptual development in the text. Another characteristic of e-textbooks that can cut both ways is the ability to link to networking sites where they can exchange thoughts about the mathematics and insights into each other’s difficulties. While that can be very beneficial, there is also the danger that these students will be tempted by the distraction of social media that is equally close at hand.
As Daniel and Willingham point out, online textbooks are easily corrected and updated. For the authors of such a text, that means that the job of working on the book is an ongoing task that is never completed, working against the ability to keep the cost down.
The bottom line is that we do not yet know how best to take advantage of online textbooks. Doing it right is clearly not as simple as putting the text on line and inserting links.
 David B. Daniel and Daniel T. Willingham. 2012. Electronic Textbooks: Why the Rush? Science. 335. 30 March, 2012. 1570–71.