I just recently stumbled across your blog post “How many US public libraries have actually closed?” and you mentioned in it that you’d appreciate any hard data on the issue.

Over the last month I’ve been doing a lot of library data analysis over at my blog ‘Library Data’ (library.tumblr.com), and your quest to find this data intrigued me so I’ve started working with the raw data from IMLS.

Here’s my first post on the topic:
http://library.tumblr.com/post/18374222475/preliminary-closing-data-i-recently-came-across-a

In that data set we’re looking at ~17000 libraries + branches, and IMLS does record official ‘closings’ rather than ‘temporary closings’, so those number should be close to what you where looking for.

Here’s the actually data:
year total_closings branch_closings
2008 101 51
2009 99 52

Later this week I’m going to try to see if I can coax numbers going further back than 2008 from the data (I think this should be possible). I actually have a more ambitious goal of looking into whether or not logistical regression can be used to predict the likelihood that a library will close in the next 5 years or so, or at least gain insight into which trends are most predictive of impending closing.

I hope you enjoy!
—–
And my response:
Thanks. I’ve bookmarked the post (and subscribed to the feed).

Those figures (ab. 0.5% per year) seem high, and I suspect that they don’t account for reopening–e.g., the Gilroy libraries, which were closed for about 18 months, and a similar situation in Oregon. The closure is “permanent” (that is, at the time of closing, there’s no direct path to reopening) but not lasting (people in a community that itself is not shutting down tend to strive hard to reopen a closed library). But it is another data source.

I should note that, as far as I can tell, the total number of library systems is growing over the years (albeit not the number of libraries per capita, a very different issue). Putting all these things together in a meaningful way continues to be an interesting issue.
–and, a few minutes later—

Now that I think about it: While I haven’t yet started trying to deal with the IMLS tables directly, one way to get some handle on actual long-term closures would be to get a *list* of the libraries reported as closed in a given year, then do some searching to see whether those libraries have reopened, say, two or three years later. (So, right now, 2008 closures would be ripe for re-investigation, and 2009 closures might be interesting.)

Hmm…

Bob,

I agree that the financial situation of too many public libraries is unfortunate. I think the apparently-false notion that libraries are shutting down all over the place hurts in two ways: First, it’s not true, and plays into the hands of “futurists” and libertarians who would *like* it to be true. Second, it obscures the real issues–which include the love most people have for public libraries and the lack of connection between that love (which sometimes prevents closing of branches that really should be closed) and adequate financial support for libraries to be the best they can be.]

walt

I did a frequency distribution on STATSTRU for FY 2009. I can send you the gory details separately if you send me your email address but here are what I think are the highlights in terms of your question and analysis:

* It looks like no libraries closed.
* 9,216 (99%) — of the libraries had no change in status.
* 5 (0.05%) — absorbed by another entity
* 1 merged with another entity.
* 18 “add an existing [library] or Outlet not previously reported.”

As I mentioned up thread, I am not familiar with the characteristics of this variable and if you wanted to do a more rigorous study, I would like to examine the variable over time and with more care.
–

All that said, there is a larger point that we should not lose sight of. I believe you are correct that the number of library buildings has not fallen and, in fact, it looks like they have increased from what I see. However, there is a argument to be made that in some cases those buildings are being “hollowed out” to borrow a term. Use of public libraries appears to be going up from the best available evidence but there are many reports of staff layoffs and declining budgets. The data lag but public libraries in states I am familiar with are taking a major hit in funding.

As a result, I don’t believe all is happy in library land with the state of public libraries at least and the national-level data we have are not reflecting it.If all this is true, the buildings stand but some number have smaller staffs and aging collections so service likely would be declining. That is what I mean by “hollowed out.”

A related question is: does the purchasing power of library budgets keep up with inflation? Last time I examined the question for public libraries, my conclusion was that over all, they had but not all libraries’ budgets had. That was through FY 2008 data. From my examination of the Gerould-ARL data from about 5 years ago, I found that the handful of (now) ARL libraries for which we have continuous financial data from 1913, had also. These observations are from memory.

Bob

Yes, Blake is a treasure, even as I argue with him over slant and general opinion. This blog (and Cites & Insights and my personal site, waltcrawford.name) is on LISHost, and I do look at LISNews every day.

So I need to look at your source, I need to do some more thinking, and then I need to decide whether this is an article worth submitting to AmLib or whether it fits in Cites & Insights. Decisions, decisions. This time, though, good decisions: If the original question has taken on an unexpected life, it’s for good reason.

thank you for your kind words.

In your honor, I have a red I in my hand about which …well, Thurber said it best: “It’s a naive domestic Burgundy without any breeding, but I think you’ll be amused by its presumption.”

I find the missed opportunities to use the abundant available data to aid the cause of libraries and the broader library function disheartening.

I was serious in what I said about 7th grade arithmetic. My argument would probably begin with my experience being asked on occasion to review articles involving data submitted to journals. In almost all cases…I’d be inclined to say “all” but maybe not…of these articles I have reviewed have confused techniques in two great domains of Statistics (capital “S”): Inferential and Descriptive.

True, the techniques overlap to an extent but to see tests of significance (t-tests are an example) on population data shows a fundamental misunderstanding of one of the major dichotomies in analytical techniques appropriate to data of varying provenance. Think Statistics 101. Bravura perhaps but with no understanding of what is going on so it may be impressive to some but meaningless.

For example, I showed a dissertation proposal for a Ph.D. candidate of in a library school to the best Statistician I knew–math stat guy–and he called it a “parody of Statistics.” I infer then, that a central analytic technique used in analysis of data in the library field is often (?) invalid.

Your analysis was appropriate to the problem, clever, and got at a deeper point. And compelling.

FWIW: Blake is a treasure and LISNews is a part of every day. I bet you agree.

Bob

I do recognize your name, with considerable respect; you’re among those who have taken statistics and analysis seriously and avoided overstatement.

No apologies are necessary. I am not so sure that a sense of libraries, a good handle on 7th grade arithmetic, and an acquaintance with spreadsheets does more good for libraries and for librarians than most of the published analytic literature.

However, that is probably a conversation best handled with a beer in hand.

Bob

I’m not sure I’m all that skilled at working with data–I tend to stick with fairly low-level analysis, and you’ll see no T-squared (or whatever) reporting in my book or articles–but I am numerate and interested in this stuff, and I have considerable interest in rigorous approaches and real-world significance. Now, if I could find funding for my 100% ongoing survey of public library presence in social networks…

There is, in fact, a variable in this excellent series which takes account of changes in library entities: “STATSTRU.” If I may suggest, you might look at the longitudinal dataset I recompiled from this public library series when I was at NCLIS and have kept up to date since then:
http://drdata.lrs.org/pldf3.

The list of variables (see the box on the right) includes STATSTRU and you can see the STATSTRU codes have varied a bit over the years but could be used to get a better handle on your question although I think you probably have it about right.

You can see from the codes that libraries close, merge, separate again, are reconfigured…and on and on. I think a frequency of these codes by year would be pretty straightforward. Whether trends are easy to pick out is another question. I have never used the variable so I am not familiar with its characteristics over time.

On the question of library data.

In fact, the library world has a number of good series but lacks a critical mass of people who are skilled in working with data. We don’t use what we have. I suspect it is easier to complain about how bad things are than work with what existing data series we have. From such work we could learn from them what we can about libraries and to improve both those libraries and the data we have on them..

Bob

While not precisely analogous situations, this parallels in some ways the public library experience during the Great Depression. In the face of economic chaos far worse than that we presently face, American towns and cities persisted in opening new libraries. See the “The American Library and the Great Depression” article on my web page.

Charley