There are two traditional uses that national-level data collected on libraries have been used for by working librarians: budget justification and supporting decisions – and each relies on finding libraries similar to the target library. We have expanded and embedded the major method for finding like libraries in the Normative Data Project (NDP), are examining other methods, and have implemented an entirely new method using Census data. One can now find like libraries by demographics of a population surrounding a library. This article will review these features of the NDP.

How do librarians use information on similar libraries? By seeing if other similar libraries appear to have more or fewer resources or are doing more with them. And if so, why? What can we do to improve? Another use of identifying similar libraries is the directory function: identifying libraries with similar situations and getting in touch with them to learn what they are doing.

Budget justification is much on people’s minds these days, having been through a time of budget constriction, and has led to a variety of new directions in thought on library data because libraries are expensive, and those expenses are obvious to funding boards, and libraries' benefits are diffuse and hard to measure. Accountability is on our minds. Though it appears that the budgetary clouds are parting for most public libraries, this requirement to justify budgets is not likely to go away, and work has taken place on several fronts to assess library benefits directly. Discussion of this question will have to await another time, but it must be said that this area of investigation is not yet mature. However, the work of Glen Holt and Donald Elliott on their method of calculating cost-benefit analysis seems to be yielding useful information, and this method has been repeated so the bugs should be pretty well worked out. A book due to be published in the fall should present the method in detail.[i]  

In public libraries, identifying “libraries like mine” has traditionally focused on per capitas, that is, comparison of a set of libraries, particularly whole states, by a given variable, per population served. “Population served” is a central data element in the public library world and refers to the number of people in the “legal service area,” which is a political concept.

Probably the most sought after numbers from the yearly publication of the NCES data appear in the annual E.D. TAB, the latest being E.D. TAB: Public Libraries in the United States: Fiscal Year 2003. These numbers are those found in its Appendices, namely the Public Library State Ranking Tables, with each of the states, plus the District of Columbia, ranked by (now) 22 ratios, such as number of library visits per capita, and so on.

In FY2002 and FY2003, the publication of the ED TABs was delayed by internal processes in NCES, and there was a good bit of discomfort, particularly with state librarians. As a result, preliminary computations were done based on the same data used by NCES and made available to those interested. In FY2002, these preliminary tables—with added analysis appeared on the Web site of the National Commission on Libraries and Information Science. In FY2003, similar preliminary tables were added to the Normative Data Project (NDP) Web site.

In working on the Normative Data Project, we have sought, among other things, to expand this ability for public librarians to identify libraries like theirs. We have expanded the current method and also developed new ways to find like libraries.

NCES reports in the ED TAB the state rank tables on 22 ratios for the states, while the NDP reports the 22 ratios in the state rank tables for all library systems, thus allowing detailed examination of a set of like libraries. By creating a view of the data of a set of target libraries, the librarian would be able to quickly do analysis on these ratios for budget reports or annual reports.

But, why do we only focus on per capitas, as reasonable as that is? Why not ratios per dollar expended, for example? They would seem as reasonable as per capitas. We are investigating several such ratios that may be included in the NDP in the future. Imagine finding libraries with similar per capitas (say, circulations per capita) and then being able to compare these libraries by circulations per dollar expended.

The Normative Data Project includes the two estimates of population served from the public library data from the National Center for Education Statistics (NCES) but also includes default “market area” estimates. The notion of market areas was developed by Christie (Christine M.) Koontz in her Library Facility Siting and Location Handbook (Greenwood, 1997). Christie is also director of the GeoLib project, which is a partner in the NDP.

The notion of library market areas is more sophisticated than that of legal service areas because legal service areas presume that users of a library are found within a political boundary, while in reality, users of a library may well come from another jurisdiction and, for example, use the library near where they work or shop, not where they live. In addition, legal service areas are for systems, not branches, because NCES data are primarily for systems. Given that some systems have many branches, it may be that a local branch in a system wants to find similar branches in other large systems, but there is now no systematic method to do this outside of the NDP. Although there are no NCES ratios available for these branches, there is a wealth of information on the characteristics of the people who live near each of the branches from the US Census.

There are about 17,000 “stationary outlets,” that is, branches that are not bookmobiles, in the FY2003 data. In order to create market areas for all of them, a default market area was set up by including any Census block within one mile of any branch in this default market area. A mile is probably too much in New York and too little in Wyoming, but it has the advantage of being quick, and we have a method to modify these defaults to more accurate market areas.

When we have more libraries that have corrected their market areas, making demographic comparisons using Census data will be much improved. Until then, well, it is pretty spectacular as it is, as a result of a new set of data in the NDP: a view of Census data that now allows one to quickly narrow down the data on branches of libraries contributing circulation and holdings data by demographic variables. In effect, these contributing libraries are indexed by all the demographic variables for any branch in the set of contributing libraries.

In the picture from the "Contributing Libraries Like Mine" screen, we are ready to make a selection on the Census variable reporting the percent of population for a given library outlet market area with an education below 9th grade. If your library has, say 15% in this category, you would select "10% to less than 20%," and all the libraries in this particular sample with a population with a 9th grade education outside this range would drop out, and you would be left with libraries like yours in this variable. By selecting other variables such as age, income, and so forth, you can select like libraries to quite a narrow set of criteria. In this case, using data from libraries contributing circulation data to the NDP would allow one to see what other like libraries are circulating. If one used these criteria on all libraries in the US, one might compare them on other variables. For instance, do they get as many visits per capita? What about their income per capita? Is it higher or lower than your library? Can we use this information in funding or grant requests? Of course, what often happens is that librarians at similar libraries call each other to swap information.

A library’s public is an important determinant of how a library is used, and having a handle on the characteristics of the population means we are a step closer to understanding how libraries are used. More enhancements to the NDP in the near future will move us even closer.