How you blog and confidence levels

Update: Please note the comment from Sarah Lovato. I do not, in fact, question that the sampling done yields “5% error margin and 95% confidence level” as defined by standard sampling statistical measures. I do question–and have questioned elsewhere–the congruence of such “error margins” and “confidence levels” when used for heterogeneous populations for which bell-curve distribution may not be meaningful.


“The bunless librarian” posted “How do you blog?” yesterday.

She “decided to determine which blogging platforms are the most popular among librarian bloggers”–and did so by taking a random selection of 275 of the 780 blogs listed in the LISZen wiki. She says that yields “results with a 5% error margin and a 95% confidence level.”

Her results? Blogger “is by far the platform of choice for the general population, followed distantly by WordPress and even more distantly by Typepad.” Specifically, she found 150 blogs using Blogger, 70 using WordPress, 16 using Typepad and 12 using Movable Type, 6 using LiveJournal and a few other numbers.

95% confident, 100% wrong…

The problem here is that this research has already been done (as of last summer)–but using 100% of a population of active liblogs with at least nominal visibility. And that 100% sample shows very different figures.

Here’s what I report in Chapter 1 of The Liblog Landscape 2007-2008, based on the 511 of 607 blogs for which the blogging platform was clearly identifiable:

  • WordPress (platform and software): 230 blogs or 45%
  • Blogger: 222 blogs or 43%
  • TypePad: 35 or 7%
  • MovableType: 18 or 3.5%
  • LiveJournal: 6 or 1%

[Note: The percentages in the book are all lower because they’re stated as percentages of the full 607 blogs, including the 96 for which I wasn’t sure of the blogging software or it wasn’t one of the “majors.”]

Basically, it’s a tie between WordPress (not just WordPress.com, as many–probably most–WordPress users have their own domains) and Blogger. I’ve seen further migration toward WordPress software, as experienced bloggers tune up their blogs; I’ve never seen anyone migrate from WordPress to Blogger.

…or not

On the other hand, it’s doubtless true that Blogger is the platform of choice for “nonce blogs”–ones started to fulfill class requirements, 23-things programs, or other cases where continued blogging is unlikely.

Shifts from 2006

Of the 184 blogs in my 2006 study that are still around, 50% used Blogger, 24% used WordPress, 9% used MovableType, 6% used TypePad back in 2006. Among that group, 42% still use Blogger (and 32% use WordPress).

But the 2006 study deliberately chose “midrange” blogs. Of the 327 blogs in the 2008 study (with clearly evident blog platforms) that weren’t in the 2006 study, 170–52%–use WordPress and 143–44%–use Blogger.

Here’s the thing…

Sarah Lovato, the Bunless Librarian, didn’t know I’d done a much broader study already. That’s not surprising: To date, fewer than 50 copies of The Liblog Landscape have been purchased. (If you’re tracking, that means it’s now about 8% probable that I’ll continue with this ongoing study.) I have no particular reason to believe I’m on Lovato’s radar at all, or that I should be.

I didn’t feel that it was reasonable to just release the results of a couple hundred hours of work as a gimme, given the lack of sponsorship, a full-time job, independent wealth and unlimited time or energy. I still don’t. It’s not particularly a secret that this book is available (several other bloggers have commented on it, a pleasant change from the near-universal tree-falling-in-a-forest reception for the two library blogging books), but it’s not widely known or read.

Yes, I know, a 95% confidence level only means that, for 20 samples of the same size, 19 of them will have results within 5% of these results. And if the sample is taken with no regard for all of the long-abandoned blogs in LISZen (and every other blog directory), blogs that tend to stay around forever if they’re on Blogger, then I wouldn’t be surprised if the results were similar to those in the post.

So maybe I should revise what I’m claiming here. I’m claiming that library people who blog on an ongoing basis are at least as likely to use WordPress as Blogger. And I guess that’s a different claim.

[Sigh. I’m torn between the desire to promulgate the results of what I regard as very good research, and the desire to scratch together a decent living and see some rewards–monetary or otherwise–for that work or the other work I could be doing in the same time. Once the price for the book went up to a price that’s still as low as, or lower than, most trade paperbacks in the library field, total sales went from an average of two every three days to a total of one in five days.]

3 Responses to “How you blog and confidence levels”

  1. Sarah Lovato Says:

    Mr. Crawford:

    The Bunless Librarian here. I’m flattered that someone of your caliber has spent his time dismantling a survey of mine, even if you were less than gentle about it.

    I could have saved you the effort of writing this post had you simply waited for me to reply to the comment you left on my blog.

    Excerpts from my reply,

    “The discrepancy between the popularity of Blogger and WordPress is interesting. Blogger coming out ahead in my survey is most likely the result of LISZEN as my population . . . . As many of the blogs listed in LISZEN are now defunct, it may be better categorized as an archive. . . .

    Curious about Blogger’s dominance on LISZEN, I compared my sample to “Blogs in The Liblog Landscape.” 131 (nearly half of my sample) do not appear on your list (probably because they are defunct or inactive) . . . . the blogs that do not appear in your list lean heavily toward Blogger. . .

    I knew that many blogs were defunct or inactive, but chose to include them because my information does not reflect “current” blogs, but blogs that are or have been authored by librarians. . . .”

    http://thebunlesslibrarian.blogspot.com/2009/01/how-do-you-blog.html

    “95% confident, 100% wrong…” This, of course, is assuming that only your work on blogging statistics is valid. While I respect the time and effort you put into your work, as I demonstrated in my reply, you were measuring apples and I was measuring oranges, which accounts for the discrepancies in the results. I must stand by my 95% confidence level and my results overall. Though I do recognize that more clarity about what exactly I was measuring may have avoided this conversation. In retrospect, I would have titled my post “How Do You Blog (Or Perhaps More Aptly, Did You Blog?).”

    Mr. Crawford, to date, I have 68 readers of my blog, 35 (a generous estimation) of which read my survey results. That is 35 people who would have read my blog and perhaps thought, through no intention of my own, “hmmmm . . . these number’s are different from those found in Walt’s book-length study.” I would have happily written a second post clarifying the difference between your work and my survey had I known the passion behind your short comment would bring us here. I did try to accomplish that clarity in my very long reply to your comment, which was unfortunately posted after “how you blog and confidence levels” appeared here.

    You have hundreds of subscribers through Google Reader, to say nothing of other feed readers. My post, which never claimed to contradict or compare to your work, was read by 35 people. This post, in which you seem to take far too much pleasure in pulverizing my oranges and credibility with your apples, was read by hundreds, maybe thousands.

    I would like to challenge you to help repair any damage that may have been done to my reputation by this post. I ask that you post this reply to your blog, in its entirety, so that I may defend my results to those who have read this post. I will gladly post it on my own blog, but as I have stated, my audience is a humble few compared to yours.

    –Sarah Lovato

  2. walt Says:

    Sarah (or Ms. Lovato, if you prefer–but I’m “Walt,” not Mr. Crawford),

    As you can see, I have no intention of truncating or editing your reply–and, in fact, have modified my post to highlight your reply.

    I’ll admit that I didn’t (and don’t) feel I was taking “far too much pleasure in pulverizing” your work. I did feel it was appropriate to point out that the conclusion as stated, that Blogger is by far the most popular platform for “librarians who blog,” is probably not true if “who blog” is defined as “who currently blog on an ongoing basis.”

    If I was going out of my way to trash your work, I most certainly would not have included these paragraphs:

    “Yes, I know, a 95% confidence level only means that, for 20 samples of the same size, 19 of them will have results within 5% of these results. And if the sample is taken with no regard for all of the long-abandoned blogs in LISZen (and every other blog directory), blogs that tend to stay around forever if they’re on Blogger, then I wouldn’t be surprised if the results were similar to those in the post.

    So maybe I should revise what I’m claiming here. I’m claiming that library people who blog on an ongoing basis are at least as likely to use WordPress as Blogger. And I guess that’s a different claim.”

    The title of the post was “How you blog and confidence levels.” It linked directly to your post. I agreed that your results were what they were–that, given your methodology and definitions, the results made sense.

    Your excerpted response doesn’t include the first paragraph of your comment:

    “Where’s the excitement in just reporting someone else’s data? It’s so much more fun to craft a pie chart instead of just inserting a link.”

  3. Sarah Lovato Says:

    Walt,

    Thank you for pointing your readers to my response. As I said, I recognize that more clearly outlining what blogs I included in my statistics and an altered title may have saved me this conversation. I am planning a new post, “How Do You Blog? 2.0,” which will clarify my intent for those of my readers who do not read the comments on my blog.

    Although I recognize that your post was overall, simply an analysis and comparison, your fourth paragraph, which is in bold, in a larger font, and follows your recap of my method and results, states “95% confident, 100% wrong. . .” While, yes, other text is similarly bold and in a large font, I felt this proclamation was misleading, despite the fact that you elaborate later in your post. As we are both sadly aware, despite all the time and effort we put into writing our blogs, most readers simply skim the content or begin reading a post curious of its content, only to never finish it. I felt compelled to respond to ensure that “95% confident, 100% wrong. . .” was not the last impression left on any of your readers.

    Again, thank you for pointing your readers to my response. It is greatly appreciated. I have similarly added an update to my post, which points to your response so that, despite how it may have felt on this end, it is clear that you were not, in fact, “taking far too much pleasure in pulverizing my oranges and credibility with your apples.”

    Happy Blogging,

    Sarah
    (Please, call me Sarah)


This blog is protected by dr Dave\\\\\\\'s Spam Karma 2: 104740 Spams eaten and counting...