Normative Data Project for Libraries

NDP Definitions

There are four sources of data currently in the NDP: the NCES/NCLIS data, data from the member libraries' ILS catalogs and their ILS circulation records, and from GeoLib.

The latter integrates Census data, geographic cata, and data on library outlets' market areas. Items in that database are defined there through the NDP interface. The remaining three are discussed here.

1. NCES/NCLIS data.

The US National Center for Education Statistics (NCES) began collecting a series of annual data from public libraries in the US and territories. These data are most useful for state and national totals because of the fact that these data have imputations for non-response. That is, if a library reports but does not fill in a given data item, NCES estimates what that number would have been, had the library responded. Imputation for non-response is a recognized technique for handling problems with estimating such as non-response and a necessary technique when the goal is national or state estimates--the primary use of the NCES data.

However, NCES in its Compare Public Library tool uses non-imputed data for those comparisons so that imputed data have been purged. Actual data as reported are used for those comparisons. Given that the NDP will also be used for comparisons, it uses a dataset from the US National Commission on Libraries and Information Science (NCLIS) that also has the imputations removed. NCES does not make the unimputed data available but its excellent documentation gives the means to remove these created variables. Documentation on this dataset, the Enhanced Longitudinal Public Library Data File is available on the NCLIS site. Only two years of this set of data are currently available here although all years are available there in various formats.

Given the close relationship between the two datasets (NCES and NCLIS) most variables and definitions are the same. The FY 2002 E.D. TAB Public Libraries in the United States: Fiscal Year 2002 (PDF) has information on the variables beginning on page 128.

However, the Library Research Service has the most useful and accessible definitions.

You will also notice that these data are system-level data. While the NDP has data on individual library outlets, the NCES/NCLIS series collects data from the various outlets and presents summary data from the systems. As a result, outlets that are part of larger systems will have no NCES/NCLIS data. There is an NCES outlet file but it mostly contains address information. One data item: square footage, is in that file.

Summary statistics from the variables in the NCES/NCLIS data for FY 2002 are available for comparison (PDF) with the NDP member libraries.

2. NDP Circulation Measures

Total Libraries
Total number of libraries reporting for all years. This number will be larger than the total number of libraries reporting in the CATALOG MEASURES (below) because more libraries circulate materials than maintain catalogs.

As measured by the ILS. This number, as well as renewals, holds, and the sum of checkouts and renewals, may or may not agree with published reports because each library may circulate materials outside of the ILS, for instance, for short times or reserves.

As measured by the ILS

As measured by the ILS

Checkout and Renewals
The sum of these two figures.

Total Copies
The sum of titles held times the number of copies of each title. Some may refer to copies as items or as the phyiscal inventory itself.

Copy Turnover Ratio
The average number of times each item in the collection circulated. Note that this ratio is often calculated by dividing volumes held by the number of circulations because volumes held is the number reported in most library surveys. To confuse this issue, there has historically been an often vague distinction drawn between physical and bibliographic volumes. That is, one library may bind a bibliographic volume of a magazine in two physical volumes while another library may bind that same bibliographic volume in one physical volume, thus clouding comparisions. The NDP database records titles and copies but not volumes, thus comparisons between two libraries are more accurate and of uncertain comparability with other turnover measures at this time.

Summary statistics for these data will be calculated and made available.

3. NDP Catalog Measures

Total Libraries
Total libraries reporting. Note that this total will be smaller than the total given for circulation measures because more libraries circulate books than maintain card catalogs.

Unique Titles by Library
A count of titles held by each library.

Total Copies
The sum of titles held times the number of copies of each title.

Average Copies per Title
Ratio of Total Copies divided by Unique Titles for each library

Percentage of Titles to Total Titles
Ratio of Titles in a given category to Unique Titles by Library. For instance, one could see what percent of the collection is in English or in the Dewey class range 300-309.

Publication Year
Mean year of publication of the title. This number comes from local cataloging.

There are anomalous data reported in various library's catalogs for this variable. All meeting the criteria where the year is below 1000 or above 2005 are coded to "unknown."

Average Price
This number comes from the catalog record. At this time, this number contains large anomalies as is discussed in detail elsewhere. We are evaluating this variable currently.

Total Collection Value
This is the sum of Item Prices. That is, Item Price times number of copies.

Checkouts and Renewals
The sum of Checkouts and Renewals as measured by the ILS.

Date Title [record] Created
This is the date stamp from the ILS for when the title was first entered into the ILS.