The Shatzkin Files

Much-trumpeted survey proves the opposite of what the surveyors seem to think it does

Tweet about this on TwitterShare on LinkedInShare on FacebookShare on Google+Share on RedditShare on StumbleUponEmail this to someone

Do library eborrowers also buy ebooks?

Well, stop the presses. OverDrive, the leading aggregator providing libaries with ebooks, and Library Journal* have done research that proves that they do. (*My error hereby corrected: American Library Association, not Library Journal.)

The survey results are interpreted as evidence that the big publishers are making a terrible mistake being cautious about making ebooks available for library lending. And it is being reported that way. By one outlet after another, although one made the point that the publishers aren’t listening.

Perhaps the publishers aren’t changing their policies because they actually are listening. In fact, the survey proves that caution makes sense. Here are the followup questions.

“What makes you decide to buy an ebook rather than borrow one? Might it be that you’re buying the ebooks that are not available to you through the library?”

I’d offer two points to ponder.

One: what if the survey had said “we have found no overlap at all! The people who borrow ebooks from libraries never buy an ebook. They only borrow.”

If that were the case, would there be any reason at all not to sell ebooks to libraries? No! There would be no potential for sales cannibalization among that audience if borrowers and buyers did not overlap.

Two: I started reading ebooks on a Palm Pilot in 1999. Between 1999 and 2007, if you’d asked me, I would have had to say I still read print books as well as digital. Why? Because a lot of what I wanted to read wasn’t available in digital. So if I wanted to read it, I had to read the print.

Then in 2007, the Kindle came along and, with some helpful pressure from Amazon, publishers routinely made their most commercial titles available as ebooks. So I wasn’t compelled to read print anymore. And, in fact, I have read only one or two printed books since then. (I remember one clearly but I think there’s another one too.)

What if the book purchasers among the library ebook borrowers have precisely the same motivation? What if they’re buying some ebooks because those aren’t available in the library?

So I’d say “thanks for the information” and for evidence from an unbiased source that publishers are entirely correct to be wary and careful about making ebooks available for library lending.

The New York Times says this morning that Penguin has announced a second set of “experiments” with ebook lending policies. The first one was with distributor 3M; this one is with Baker & Taylor. They have chosen time-limited (1 year) licenses as their gambit. As I told the Charleston audience last week, our client Recorded Books is about to debut an ebooks-for-libraries program that will give publishers the flexibility to control four aspects of the license: in or out of catalog; price; number of loans limit (if any); length of license (if limited). Obviously, the Penguin experiment could be conducted with the RB capability, with the advantage to them that they could vary the terms by title and change them over time (always honoring deals already made, of course.) 

Doing this requires a tool set for libraries to manage their ebook collections, which Recorded Books will provide. Over time, the RB capability should enable enough experimentation to bring far more titles into the library marketplace while allowing publishers to learn what works and adjust to what almost certainly will be a changing environment.

Tweet about this on TwitterShare on LinkedInShare on FacebookShare on Google+Share on RedditShare on StumbleUponEmail this to someone

  Back to blog

  • Troy Johnson

    The survey seemed to have a rather narrow demographic — The respondents were overwhelmingly female (78%), earned more more than
    $50 thousand a year (75%), were in their 40s or older (72%), and were
    college educated (73%).

    • It doesn’t “look like America” but maybe ebook borrowers from libraries just don’t. I really have no clue.


  • Hi Mike–I think you might have actully meant the American Library Association? It was they (not Library Journal) that partnered with OverDrve on this survey.

    • So sorry for the error. Thanks for correcting it!!!

      Now I did too.


  • The fact that Overdrive commissioned the research has to make one sceptical about its findings. Ditto if a trade association for cabbage growers commissioned research on whether cabbage is a healthy food.

    • Oh, I think the findings were perfectly accurate. Or I have no reason to doubt them.

      What I think was *not* accurate was the interpretation of them. I have no idea how much influence OD had over that. Certainly the press that reported on them were not corrupted by OverDrive.


      • Sorry, I did not mean to imply any fudging of the results. But I think a book eTailer would have used different questions and obtained a result more in line with your interpretation of the Overdrive survey.

  • There are a couple of peculiar things about this survey other than those already noted. First, once I noticed that for each question the survey indicates how many respondents answered it and how many skipped it It quickly became apparent that a very large number — nearly 14% — skipped fully *half* the questions. And second, even stranger is that the very same number of respondents — 10,328 — skipped these questions. What are the odds of the exact same number of respondents skipping 9 of 18 questions in a survey of 75,385 people?

    What’s more, the answer to the survey question that prompted this post — “In a typical month, how many books (digital and physical) do you purchase at retail (including audiobooks)?” appears to have been interpolated in the PDF you link to, Mike. This survey question clearly solicits a numerical answer (e.g., none, 1 or 2, 3 to 5, more than 5), but the answer is expressed only as a distinction between buying physical books from a physical or online store or buying digital ebooks or audiobooks from an online store. (And in that respect the answers are nearly identical: about 94% for one and about 98% for the other.) Then, too, notice that this important question is flagged as OPTIONAL and was skipped by more than a *quarter* of the survey respondents.

    Sure, at the end both Overdrive and the ALA admit the survey isn’t scientific — it’s “not representative of the US population as a whole.” But it’s always annoying to review a survey and notice certain anomalies — especially when they suggest a certain massaging of the results.

    • Thanks for this analysis. I was so stunned by what I saw as the misinterpretation of the bottom line result that I didn’t look deeper for other anomalies. Once again, I’m proud to have one of the smartest and best-informed comment strings in the publishing blogosphere. I always learn a lot from you and my other readers. I often think about the fact that most of my traffic doesn’t tune in at the right time to derive this benefit from our intereaction.

      My off-the-top-of-my-head thought about the question set unanswered by precisely the same number of people is that it hardly seems possible. Even if they withheld those questions from a big chunk of those polled, you’d stll be surprised to see the same number of respondents for all the rest!

  • Great analysis Mike (as always):

    When I saw the report, I thought two things:

    1) Imagine that I (or anybody else like me) would rather borrow for free – any ebook they can get at the local library and if they can’t, they’ll buy it. The survey data would be 100% aligned with such a model.

    2) People who borrow books form libraries also buy books, but the survey never, never tells us if people buy the SAME books they borrowed. All it shows is that people who use the libraries frequently (i.e. they read a lot) also buy a lot of books. Surprise, surprise.

    To pick up on your point above: what we need is the data for a Venn diagram.One circle would be library users and one circle would eb book buyers. The folks in the overlapping section of the two circles would be those who willingly buy books (especially if they are not available for loan) while those in the non-overlapping sectiona would be (1) library patrons who never buy books (they only read for free) and (2) those who always buy the book and the (i.e. because using Overdrive software is just too clumsy an experience or because they don’t want to spend tax payer money, etc.).

    If there were no overlap between the circles, then we would have the scenario you – hypothetically – describe above (library buyers never buy books), but if we have 90% overlap, then we basically have a case of huge, huge cannnibalization.

    The problem is that Overdrive et al have no incentive to collect THAT kind of data (to echo the point of another commenter).

    • Your point number 1 is exactly the way I see it.

      Has what I said been said by anybody else yet? Kind of amazing to me that this isn’t obvious to many people.


      • spin, spin, spin

        or because most people don’t have much training in stastics or Baysian logic

  • I am a librarian and I argued something similar earlier this week. There is also the commodification effect that needs to be considered. Ebooks have a potential for commodification more than print items, which means they are fungible, meaning people dont care where they get them as long as they can. see my post at

    • Joseph, a deep link to the post you want us to see would be much more helpful than a link to the home page of your site, which this appears to be.

  • Pingback: WRITING ON THE ETHER: Eye of the Turkey | Jane Friedman()

  • oysterfarm

    Couple of points: the argument of the survey results is simply that people both buy and borrow. In some cases, they may buy what they have borrowed and, in many cases not. It is the same with printed books – library borrowers buy books, whether electronic or print. The important point is that they are in the market place, buying. What they buy is irrelevant.

    On the point about the same number of respondents not completing: this would only be remarkable if they were the same people.

    • I think it is reasonable to assume that most people who buy and borrow would tend to be less likely to buy what they could borrow. I can’t prove it, but it seems intuitively correct. So what they buy is totally relevant to the publisher who is trying to sell a particular book.

      And I know enough about statistics to know that hitting the same number raises a flag, given the range of realistic possibilities. If they were the *same* people, that could only happen by some outside force. The laws of probability would definitely prohibit it.


  • Pingback: Reading about eReading this week 11/26/2012 « Allegany County Library System Director's Notes()

  • Pingback: WRITING ON THE ETHER: Springtime for Librarians()