I did an in depth review of the research for a Human Computer Interaction class in college a couple years back before the Taylor and Francis study and the research then generally agreed with the idea that reading a print book was the best for comprehension (I tried quickly searching for the paper to pull sources, but no luck). However that difference appeared to be mostly related to form factor and not necessarily from the screen itself. Studies that used tablets showed smaller gaps in performance than studies that used computers. Studies that used e-ink e-readers showed almost no difference in comprehension. I was not able to find a single study that compared more than 2 or 3 different devices using the same methodology. If I were to design a study I would want to see a full range of tests including print books, e-ink e-readers, tablets with full color screens, traditional computer monitors, and e-ink monitors. Hopefully then you could better isolate the cause of the lower comprehension.

My own hypothesis (I have a comp sci degree, so take this psychology explanation with a grain of salt) is that comprehension is more dependent on the person's approach to the technology rather than the technology itself. Books are single purpose devices. When you have a book in your hand your brain knows it is time to focus on reading. When you sit down at a computer, the brain doesn't know what to expect or to focus on. It is similar to other advice about training your brain to expect certain activities such as reserving your bedroom only for sleeping or to have a dedicated home office if you work remotely.