I Geek Information Accuracy

The second phase of OCLC’s national library advocacy project debuted last year as the “GeekTheLibrary” campaign. The campaign is cool, chic, hip, flashy, geeky, and so on. Pretty ambitious to try to coin new slang! For sure I’m not a good judge of communication campaigns, but I wish them success on this one.

When it comes to data, though, these advocacy campaigns can be clueless. They really need to kick it up a notch, clean up their acts, circle back, and cut out the BS (bogus stuff). Which is to say they need to think more critically about the claims they make.1 The results will be worth the effort. As it is, the GeekTheLibrary site repeats the same ol’ same ol’ I’ve talked about already. Is it really so painful to just tell it like it is?  Like, tell the whole data story in a balanced and, like, intelligent way?

Anyway, let’s take a look at some “facts” GeekTheLibrary has posted. The first fact is:

More cards, less money. More Americans than ever have library cards, but many public libraries report stagnant or declining state and local funding levels.

The statement describes two more general and apparently contradictory trends: Demand for library services has increased dramatically while library funding has not kept pace. (This will ring a bell if you’re following ALA’s research briefs or read my April 2010 blog entry.)

To figure out if the statement is true we first need to know what timeline GeekTheLibrary is talking about. The past year, past several years, past decade, or longer? Weirdly, the site is devoid of citations of sources. Maybe that’s too geeky, but it helps when you’re trying to piece together an argument based on data. The library card part of their statement comes from a 2008 Harris Interactive survey which found the percentage of library card holders in the U.S. to be 68%, an increase of 5% from 2006. So we’re talking recent trends. As for the second part of the GeekTheLibrary’s statement, take a look at the chart below from the IMLS Public Libraries Survey Fiscal Year 2008. It shows that overall public library operating expenditures from 1999 to 2008 have increased steadily even when adjusted for inflation.

Chart 1.  U.S. Public Library Expenditures 1999 through 2008
        Source: IMLS, Public Libraries Survey Fiscal Year 2008

The GeekTheLibrary claim, though, is that “many” libraries have had stagnant or decreasing budgets. This wording is like the familiar tactic in advertisements inviting customers to “Save up to 50%!” The idea is to get the audience to infer well beyond the statement. To interpret “up to 50%” as “definitely 50%” and “many” as “a lot” or “a high number.” For the claim at hand, we must ask the question, “How many?” And then determine if the answer really is a lot.

ALA’s Public Library Funding & Technology Access studies have recent data on libraries reporting decreased or unchanged budgets as shown here:

U.S. Public Libraries Reporting Decreased or Unchanged Budgets
Data Sources:  ALA, Libraries Connect Communities: 2009-2010 and        ALA, Libraries Connect Communities 3: 2008–2009

In terms of library spending we see a longer term trend of steady growth interrupted by the severe economic recession. Is the number of libraries reporting unchanged or decreased budgets “a lot?” That is hard to tell given the current economic environment where avoiding budget cuts is nearly a miracle. Even in these calamitous times, in 2010 68% of libraries reported either increased or sustained budgets, as did 85.8% in 2009. Not wonderful news, but certainly not disastrous.

Moving on, the next GeekTheLibrary fact reads:

Low Funding. Twenty-six percent of U.S. public libraries operate on less than $50,000 annually.

Taken literally, this just means that there are some really small libraries with really small budgets, probably in really small towns and villages. Nothing new or particularly profound. But the implied message is that some libraries—mainly the tiniest ones, but perhaps larger ones also—are neglected and underfunded. And that this is not a good situation.

Let’s explore this claim further. Based on 2008 IMLS data, 94% of libraries with total operating expenses of $50 thousand or less have service area populations of 5 thousand or less, and 74% (of the $50 thousand or less group) serve populations of 2 thousand or less. Chart 2 illustrates this using a scatterplot.

Chart 2.  U.S. Public Libraries with Operation Expenditures $50,000 or Less
                                 by Service Area Population   (N = 2192)
               Data Source: www.imls.gov    Click chart for larger image.

A scatterplot helps us understand how two different quantities relate to each other—in this case, how the size of a library’s operating expenditures matches the size of its community population. The title for chart 2 displays the total number of libraries depicted (“N=2192”). On subsequent charts you’ll see this number decrease since each chart depicts a smaller group of libraries.

In chart 2 the location of each ‘+’ symbol represents one library’s operating expenditures and service area population together. As indicated in the axes labels [“Operating Expenditures (thousands)” and “Population (thousands)”], the vertical placement of the ‘+’ symbol indicates operating expenditures and the horizontal placement indicates population size. For example, notice the lone ‘+’ in the upper right corner of chart 2. That library’s operating expenditures are about $43 thousand and its community population is approximately 45 thousand.

By lessening the span of the horizonal axis to, say, 20 thousand—temporarily ignoring any libraries with populations larger than that—we can cause the ‘+’ symbols to spread out some (see chart 3). But still, in both charts 2 and 3 we see that most (94%) of these libraries are bunched to the left of the 5 thousand population mark. Note in chart 3 that about 17 libraries (count them!) serve communities greater than 10 thousand population. And in chart 2 we can see that an additional 7 or 8 serve communities larger than 20 thousand.

Chart 3.  U.S. Public Libraries with Operation Expenditures $50,000 or Less
                             and Population 20,000 or Less   (N = 2178)
               Data Source: www.imls.gov    Click chart for larger image.

Considering just these 25 or so libraries, we might agree with the GeekThe- Library claim that these are underfunded relative to their community size. To strengthen this argument, we’d need some further information about what appropriate spending levels and collection sizes are for libraries serving communities of 10 thousand to 50 thousand. Plus, we’d need to look for unique circumstances that might otherwise justify low funding at these libraries.

For the remaining group, and especially those libraries serving populations of 5 thousand or less, there’s little reason to presume funding is inordinately low. Chart 4 shows the distribution of this smaller group. (Note “N=2062” in the chart title.) About 400 of these libraries serve communities larger than 2 thousand residents. The rest (representing 74% of all libraries with expenditures of $50 thousand or less) serve very small communities. At the left side of chart 4 the symbols are denser, forming a diagonal “swarm” upward and rightward. We can see this more closely in chart 5.

Chart 4.  U.S. Public Libraries with Operation Expenditures $50,000 or Less
                             and Population 5,000 or Less   (N = 2062)
               Data Source: www.imls.gov    Click chart for larger image.

Chart 5.  U.S. Public Libraries with Operation Expenditures $50,000 or Less
                             and Population 2,000 or Less   (N = 1626)
               Data Source: www.imls.gov    Click chart for larger image.

Chart 5 shows libraries serving communities of 2,000 population or less. The chart is a close-up of the “swarm” seen in chart 4, although here the smaller population range in the horizontal axis spreads the symbols out sideways. Due to the “swarminess” (the statistical term is “variation”) it’s a little tricky to come up with summary statements describing this group. Still, the pattern is not a complete hodge-podge (it’s not completely random). In fact, the points are clustered around the diagonal line drawn on the figure. And they are more sparse in the upper left and lower right corners. The line suggests that the level of library expenditures and population match up, roughly speaking. The higher the population, the higher the expenditures. We could even say that low funding levels are “par for the course” for libraries in communities this small.

But couldn’t this whole group of libraries be underfunded? Yes, they could be. And so could all of the biggest U.S. public libraries also. But citing small library budgets as evidence of underfunding, as GeekTheLibrary does, makes no sense. There are various reasons for libraries being small, most quite legitimate. The key question is, “How much (funding, staffing, building space, technology, and so forth) is enough?”

This question leads us to one final and juicy GeekTheLibrary fact:

Get in line. Demand for computers in most libraries is so great that there is often a lengthy wait time…

Notice the quantitatively vague terms “often” and “lengthy.”2 Again, recall those “Save up to 50%” ads. The message is that because patrons (how many?) frequently (how frequently?3 ) have to wait a long time (how long?) for services, libraries (how many?) need more money (how much?) to buy more (how many?) public access computers or additional (how much?) WIFI capacity.

Good public management always requires balancing resources with need or demand. The particular problem of wait times is a quantitative one that happens to fall under the rubric of operations research. A common operations research example in public services is the distribution of emergency resources like fire fighting personnel and equipment or police patrol cars in a community. The goal is to minimize response (wait) times, provide adequate geographical coverage, and deliver ample resources to respond effectively to emergencies—without over-spending.

Providing emergency services that guarantee quick response means resources will sit idle much of the time. When the safety risk is high, having idle resources is clearly justified. In the case of less crucial public services including library public access computers, there is no justification for underutilized capacity.

Corporations know this principle well. The most relevant example is Disney World where waiting lines are an accepted part of the “experience.” Disney employs Ph.D’s in queuing theory to fine tune their operations in order to maximize throughput on rides. Success is having a sufficient rider capacity that is busy all of the time while making wait times acceptable. Kaiser Fung explains how this works in his book Numbers Rule Your World: The Hidden Influence of Probability and Statistics on Everything. He explains that for Disney’s rides and also urban highway traffic flows, it is impossible to avoid occasional long delays no matter how many resources you muster.

The goal of capacity planning is to minimize aggregate delay, that is, to satisfy as many customers as possible as often as is feasible.4 Thus, wait times do not necessarily mean there are too few public access computers (or lanes on a highway). Wait times can just as easily be an indication of good planning by library administrators.

Speaking of waiting, this lengthy (1,992 word) diatribe begs for a conclusion. So here goes: What’s up with unfounded library advocacy claims?  “So much more of this than that?” “Amazing amounts of these?” “Dismal amounts of those?”  We gotta get our geek on by being smarter.

Oh, yes, I have a confession also. I have an ulterior motive for including the scatterplots here. They are meant to illustrate how intricate data are. And how important it is to snoop around to see what is really there. Otherwise you’re liable to not know what you’re talking about.

 
—————————

1  Leo Burnett USA, the high-powered advertising firm OCLC hired for these projects, geeks critically evaluating their performance(s). They write, “We rate our creative [sic] with critical eyes and no excuses.”
2  I ignore the word “most” because we know it means somewhere between 51% and say 80%. At the high end of this range the phrase “a vast majority” applies. And for 90% and above “almost all” clicks in.
3  The project’s press kit fact sheet says waiting lines for computers are “commonplace.”
4  Or in the collectivist logic of Spock, “The needs of the many outweigh the needs of the few…”