Kindle Reading Speed Study: Badly Misunderstood

For the past week or so, blogs like ours here have been buzzing with thoughts about a study done of relative reading speeds between the Kindle, iPad, PC Monitor and Paperback Book.  The general consensus seems to have been anything from “See, eReaders are bad!” to “Look, it proves the iPad is better than the Kindle!”  This leads me to believe that a large number of people have only a very vague understanding of what this study actually means.  Let me explain.

In the actual text of the reading speed study, we are given the details of their methods.  The sample size is actually quite small, with only 32 people involved total of whom a mere 24 were included in the final data set.  Putting aside that flaw, the data gathered provided no useful information at all besides that reading on anything but a computer monitor is preferred.  For those who are talking up the slight difference in reading speed between the iPad and the Kindle, there is a note in the results that “the difference between the two devices was not statistically significant”.  For those who do not have any statistics/science background, this means that no difference can be said to exist, with any reasonable accuracy, that stem from anything but random chance.

Basically, if you were hoping for scientific evidence of which device is better, even if you judge “better” in terms of how fast you can read, there’s nothing in this recent study to help you out.  Maybe next time.

10 thoughts on “Kindle Reading Speed Study: Badly Misunderstood

  1. Any differences in reading speed may also have to do with line length and lines per page. Tweaking of these settings would affect the results. I thought reading large fonts on the kindle would be nice, but the down side is the loss of words per page and thus a reduction in reading speed.

  2. I think it’s amazing that a study with a data set of 24 people would be considered newsworthy. I’d expect to see better numbers than that in a fifth-grader’s science fair project. It’s not like it should be so hard to find subjects for a study like this — you just need people who can read, for cryin’ out loud.

  3. This is exactly why I ended up getting a DX and selling my kindle 2. I found that with the small screen (especially with slightly larger fonts) I can read noticably faster which is much more comfortable for my natural reading speed.

  4. I have Kindle DX and I used a letter 10 to 12 p. I don’t feel I am slower than with a book.
    On the contrary I would said that I am a faster reader today with the Kindle, you get very fast in turning pages. This is in my opinion the best invention in the last years. For the simplicity. The more similar to a book, is where is going to catch the eye of people that have interest in reading at a very low cost.
    The gizmo’s they add, more expensive is going to be,and less people is going to buy them. You really don’t need so much memory if you have a computer where to keep your files.
    Light, good ergonomic (button to change pages in both sides), the screen 10″ with an easy border to handle it and good contrast.

  5. Hi!
    Ok, let us suppose that they are all equal in reading speed , I think ipad will be the better choice because it give us more than reading device.
    With ipad we can read, play, work :)

  6. I have a PhD and have run a number of within-subjects studies similar in design (although not content) to this. The fact that the experiment used 24 participants in no way invalidates the results. For a within-subjects design, where everyone does every condition (Kindle, book, etc), that should be plenty to detect a reading-speed effect large enough for people to care about. In this case, a roughly 10% change was detected with little chance of a fluke result.

    Now, there may be other issues that confound the results, and presumably other people will have ideas about those, run their own studies, and report their results. Eventually, we’ll have a pretty good idea whether there are effects of device type on reading speed, and why, and probably some techniques for mitigating any real effects. I don’t think the results of a single experiment can address anything more than the existence of an effect, though.

  7. If sample size was calculated before the experiment was run, then the confidence level can be determined. If not, then we don’t know the confidence level, although the study did say it was statistically significant, it didn’t say at what level the test was done. Also, if some of the data were thrown out, it invalidates the experiment unless all the data attributed to a cause is also tossed out, even if they fit within the model. I can prove anything with random numbers if I can toss out all the outliers, sort of like the first cholesterol study, which was very defective because they tossed out the mediterranean countries. I find that MOST experiments are defective because the experimenter was not trained in experiment design nor did they consult someone who was. The most obvious example of bad experiment design is Big Pharma, who run experiments over and over until they get one result that they like and take that as the final answer. Liars, Damn Liars and Statisticians.

  8. So, they tested a small number of people who weren’t used to reading in any of the electronic forms, but probably *are* used to reading on paper, then tossed out a quarter of the results, and claimed that they had a study…

  9. @amar, you would be right, except for the time when the ipad has a dead battery and the kindle still has most of it’s charge, or if you are in bright light where the ipad screen can’t be read while the kindle can, or (since this test was with the 6″ screens) where the ipad was left behind/damaged because it is significantly bigger and heavier than the kindle, or where the choice is dedicated e-reader or nothing (due to the fact that the ipad is >4x the price)

  10. To be fair, they say in the study itself that there is no way to determine at this time which of the devices can be read faster, that the statistical difference is small enough and there are enough variables that you cannot make a definative statement. Good for them for being honest.

Leave a Reply

Your email address will not be published. Required fields are marked *