Attachment_6_-_PLS_FY2012_Datafile_Documentation.pdf

Data File Documentation Public Libraries Survey

Fiscal Year 2012 July 2014

Public Libraries Survey Fiscal Year 2012

Data File Documentation

Deanne W. Swan Justin Grimes Timothy Owens Kim A. Miller Institute of Museum and Library Services

J. Andrea Arroyo Terri Craig Suzanne Dorinski Michael Freeman Natasha Isaac Patricia O’Shea Regina Padgett Peter Schilling U.S. Census Bureau

C. Arturo Manjarrez Director, Office of Planning, Research, and Evaluation Institute of Museum and Library Services

Susan H. Hildreth Director

About the Institute of Museum and Library Services

The Institute of Museum and Library Services is the primary source of federal support for the nation’s 123,000 libraries and 35,000 museums. Our mission is to inspire libraries and museums to advance innovation, lifelong learning, and cultural and civic engagement. Our grant making, policy development, and research help libraries and museums deliver valuable services that make it possible for communities and individuals to thrive. To learn more, visit www.imls.gov and follow us on Facebook and Twitter.

As part of its mission, IMLS conducts policy research, analysis, and data collection to extend and improve the nation’s museum, library, and information services. IMLS research activities are conducted in ongoing collaboration with state library administrative agencies; national, state, and regional library and museum organizations; and other relevant agencies and organizations. IMLS research activities are designed to provide consistent, reliable, complete, and accurate indicators of the status and trends in library and museum services; and to report timely, useful, and high-quality data to Congress, the states, other policymakers, practitioners, data users, and the general public.

Contact Information

1800 M Street NW, 9th Floor Washington, DC 20036-5802

202-653-IMLS (4657)

www.imls.gov

The IMLS World Wide Web Home Page address is http://www.imls.gov.

For questions or comments, contact Research@imls.gov.

This publication is available only on the Web at http://www.imls.gov/research/

Cover Design: Ellen Arnold Losey

Suggested Citation Swan, D. W., Grimes, J., Owens, T., Miller, K., Arroyo, J., Craig, T., Dorinski, S., Freeman M., Isaac, N., O’Shea, P., Padgett, R., & Schilling, P. (2014). Data File Documentation: Public Libraries Survey: Fiscal Year 2012 (IMLS-2014–PLS-02). Institute of Museum and Library Services. Washington, DC.

Unless specifically noted, all information contained herein is in the public domain and may be used and reprinted without special permission. Citation of this source is required.

http://www.imls.gov/ http://www.facebook.com/USIMLS http://www.twitter.com/us_imls http://www.imls.gov/ http://www.imls.gov/ mailto:Research@imls.gov http://www.imls.gov/research/

Table of Contents

Page

I. Introduction

II. User's Guide

A. Survey Methodology

B. Guidelines for Processing the Data Files

Appendixes

A Record Layout for Public Library Data File, FY 2012

B Record Layout for Public Library State Summary/State Characteristics Data File, FY 2012

C Record Layout for Public Library Outlet Data File, FY 2012

D State Codes

E Libraries with No Central Outlet

F Survey Questionnaire

G Imputation Flags and Definitions for Public Library Data Files

H Item Response Rate and Total Quantity Response Rate by State by Item

I Item Response Rate and Total Quantity Response Rate by Item by State

J Frequencies of Categorical Variables on Public Library Data File

K Distributions of Continuous Variables on Public Library Data File (Suppressed)

L Frequencies of Selected Variables on State Summary/State Characteristics Data File

M Frequencies and Distributions of Selected Variables on Public Library Outlet Data File

This page is intentionally left blank.

Public Libraries Survey, Fiscal Year 2012

Data Documentation

I. Introduction

The Public Libraries Survey (PLS) is a voluntary survey conducted annually by the Institute of Museum and Library Services (IMLS). IMLS collects these data under the mandate in the Museum and Library

Services Act of 2010 (PL 111-340) as stated in SEC. 210. The U.S. Census Bureau is the data collection agent for IMLS. The Fiscal Year (FY) 2012 survey is the 24th in the series.

The data file includes all public libraries identified by state library agencies in the 50 States, the District of Columbia, and the outlying areas of Guam, the Northern Mariana Islands, Puerto Rico and the U.S. Virgin Islands. The reporting unit for the survey is the administrative entity, defined as the agency that is legally established under local or state law to provide public library service to the population of a local jurisdiction. In this survey, the term public library means an administrative entity. The administrative entity may have a single outlet or multiple outlets (single- and multiple-outlet libraries that do not have a central outlet are listed in Appendix E). The data for a multiple-outlet library are combined. The survey questionnaire is in Appendix F. For the definition of a Federal State Cooperative System (FSCS) Public Library, see item 203 in the Administrative Entity Data Element Definitions.

The FY 2012 PLS collected the following information:

State characteristics data, including the reporting period starting and ending dates, the state total population estimate, and the total unduplicated population of legal service areas for the state (see the survey questionnaire in Appendix F, items 100-103). Each state library agency reported these data on the “State Characteristics” record because they are not library-level data.

Data on each public library, such as its name and address, population of legal service area, service outlets, collections, full-time-equivalent staff, and operating revenue and expenditures (see Appendix F, items 150-651). These data were reported on the “Administrative Entity” record.

Data on each public library service outlet, such as its name and address, type, county location, metropolitan status, square footage, public service hours per year, and number of weeks a library outlet is open (see Appendix F, items 700-714). These data were reported on the “Outlet” record.

The Public Library Survey Data File is available in two versions: a public-use data file and a restricted-use data file. The public-use data file is available to all users. The public-use data file has some data suppressed, in order to protect privacy and to prevent disclosure of individual information. If there is a need for access to this suppressed information, data users should contact IMLS about procedures for obtaining access to the restricted-use data files. This document is the documentation for the public-use data file. See Confidentiality and Public- and Restricted-Use Data Files in the next section for more information.

Three public-use data files were generated from the FY 2012 PLS. These data files are provided in SAS (.sas7bdat), comma-delimited (.csv), and ASCII (.txt) formats

. These are the final data files.

1. Public Library Data File (filename: pupld12a). This file, also known as the Administrative Entity file, includes a total of 9,305 records. This file includes data for 9,294 public libraries (9,233 public libraries in the 50 states, the District of Columbia, and 61 public libraries in the outlying areas of Guam, the Northern Mariana Islands, Puerto Rico, and the U.S. Virgin Islands). The remaining 11 records are administrative entities that closed or temporarily closed for FY 2012 (STATSTRU, Comma-delimited files can be opened with Microsoft Excel or other spreadsheet programs.

Structure Change Code, is ‘03’ or ‘23’). Records for public libraries that closed for the current year are included on the file for that year only. Each library’s data consist of one record. Appendix A contains the record layout.

2. Public Library State Summary/State Characteristics Data File (pusum12a).

The data for each state or outlying area consist of one record (a total of 55 records are on the data file). Appendix B contains the record layout. No data are suppressed. The file includes:

a. State summary data. These are totals of the numeric data on the restricted-use Public Library

Data File for each state and outlying area.

b. State characteristics data. These data consist of four items reported by each state and outlying area on a “state characteristics” record: the earliest reporting period starting date and the latest reporting period ending date for their public libraries, the state population estimate, and the total unduplicated population of legal service areas in the state.

3. Public Library Outlet Data File (puout12a). This file includes a total of 17,586 total records. This file includes identifying information and a few basic data items for 17,456 public library service outlets (central, branch, bookmobile, and books-by-mail-only outlets). The file includes 17,317 outlets in the 50 states and the District of Columbia and 139 outlets in the outlying areas. The remaining 130 records are outlets that closed or temporarily closed for FY 2012 (STATSTRU, Structure Change Code, is ‘03’ or ‘23’). Beginning with the FY 2008 file, records for public libraries that closed for the current year are included on the file for that year only. The data for each outlet consist of one record.

Appendix C contains the record layout. No data are suppressed.

II. User's Guide

II. A. Survey Methodology

Survey Universe

The PLS is designed as a universe survey. The survey frame consists of 9,294 public libraries – 9,233 public libraries in the 50 states and the District of Columbia and 61 public libraries in the outlying areas of Guam, the Northern Mariana Islands, Puerto Rico, and the Virgin Islands – as identified by state library agencies.

(Public libraries in one outlying area, American Samoa, are not included in the survey frame because their state library agency has never responded to the request for participation in the survey.

Because their public libraries have not been identified, they are not included in the response rate calculations.) The survey frame (and the survey response rates in the next section) includes 154 public libraries that do not meet all the criteria in the FSCS Public Library Definition (see Appendix F, item 203 of the Administrative Entity definitions for the criteria). These libraries are included because they qualify as public libraries under state law. Military libraries that provide public library service and libraries that serve residents of institutions are not included.

Survey Response

Unit response. A total of 9,056 of the 9,294 public libraries in the survey frame responded to the

FY 2012 PLS,

for a unit response rate of 97.4 percent. Respondents to the survey are defined as public libraries for which the following data were reported: population of the legal service area and at least three of the five following items: total paid employees, total operating revenue, total operating expenditures, print materials, and total circulation. Some individual survey items, such as population of

Public libraries in one outlying areas, American Samoa, are not included in the survey frame because their state library agency has never responded to the request for participation in the survey. Because their public libraries have not been identified, they are not included in the response rate calculations.

Including Guam and Puerto Rico.

legal service area, service outlets, and type of legal basis, have a 100.0 percent response rate for their state because the state library agency provided these data for all public libraries in their state. For library unit response rates by geographic area, see the table below.

Total response. The base for calculating response rates to individual survey items is the total number of libraries in the survey frame, including unit non-respondents.

Data File and Publication Response Rates. The total response rates on the data file differ from the total response rates in the published report because the non-FSCS libraries are excluded from the publication tables. Also, the non-responding outlying areas of the Northern Mariana Islands and the Virgin Islands are included on the data file but are not included in the publication. The responding outlying area of Guam is included in both the data file and the publication. The response rates for the outlying area of Guam are not included in the national totals in the publication. The responding outlying area of Puerto Rico is included in the data file, but is excluded in the publication because response rates are very low and missing data are not imputed for outlying areas.

Library Unit Response Rates by Geographic Area

Geographic area Respondents Total units Unit response rate

Alabama 220 220 100.0

Alaska 84 84 100.0

Arizona 83 93 89.2

Arkansas 53 57 93.0

California 181 183 98.9

Colorado 114 115 99.1

Connecticut 182 194 93.8

Delaware 21 21 100.0

District of Columbia 1 1 100.0

Florida 72 79 91.1

Georgia 61 61 100.0

Guam 1 1 100.0

Hawaii 1 1 100.0

Idaho 103 103 100.0

Illinois 615 623 98.7

Indiana 236 237 99.6

Iowa 525 542 96.9

Kansas 323 328 98.5

Kentucky 119 119 100.0

Louisiana 68 68 100.0

Maine 254 266 95.5

Maryland 24 24 100.0

Massachusetts 366 370 98.9

Michigan 388 391 99.2

Minnesota 137 138 99.3

Mississippi 51 51 100.0

Missouri 149 150 99.3

Montana 82 82 100.0

Nebraska 220 268 82.1

Nevada 22 22 100.0

New Hampshire 225 230 97.8

Geographic area Respondents Total units Unit response rate

New Jersey 270 297 90.9

New Mexico 90 92 97.8

New York 756 756 100.0

North Carolina 77 77 100.0

North Dakota 82 82 100.0

Northern Mariana Isl 0 1 0.0

Ohio 251 251 100.0

Oklahoma 118 118 100.0

Oregon 129 129 100.0

Pennsylvania 454 456 99.6

Puerto Rico 18 58 31.0

Rhode Island 48 48 100.0

South Carolina 42 42 100.0

South Dakota 108 112 96.4

Tennessee 185 185 100.0

Texas 559 559 100.0

Utah 69 72 95.8

Vermont 165 182 90.7

Virgin Islands 0 1 0.0

Virginia 91 91 100.0

Washington 61 61 100.0

West Virginia 97 97 100.0

Wisconsin 382 382 100.0

Wyoming 23 23 100.0

United States 9,056 9,294 97.4

Caveats for Using these Data

The data include imputations, at the unit and item levels, for non-responding libraries. For a discussion of the imputation methodology, see the Imputation section. Comparisons to estimates prior to FY 1992 should be made with caution, as earlier estimates do not include imputations for nonresponse, and the percentage of libraries responding to a given item varied widely among the states.

State data comparisons should be made with caution because of differences in reporting periods (see Table A–1) and adherence to survey definitions. The definitions used by some states in collecting data from their public libraries may not be consistent with the PLS definitions.

The District of Columbia, although not a state, is included in this report. Special care should be used in comparing data for a city to state data. Caution should also be used in comparing Hawaii’s data to other states as all public library data are reported under one entity, the Hawaii State Public Library System.

Reporting period. The FY 2012 PLS requested data for state fiscal year 2012. In some states, the FY reporting period varies among local jurisdictions; these states are listed in the Other column in Table A–1.

However, each public library provided data for a 12-month period. Note: The FY starting date and ending date of each public library are included on the data file.

Table A–1. Reporting periods of public libraries: Fiscal Year 2012

July 2011 January 2012 through through

June 2012 December 2012 Other

Arizona Arkansas Alabama

California Colorado Alaska

Connecticut Indiana District of Columbia

Delaware Kansas Florida

Georgia Louisiana Idaho

Hawaii Minnesota Illinois

Iowa North Dakota Maine

Kentucky New Jersey Michigan

Maryland Ohio Mississippi

Massachusetts South Dakota Missouri

Montana Washington Nebraska

Nevada Wisconsin New Hampshire

New Mexico Puerto Rico New York

North Carolina Pennsylvania

Oklahoma Texas

Oregon Utah

Rhode Island Vermont

South Carolina Guam

Tennessee

Virginia

West Virginia

Wyoming 1The reporting period varies among localities for the states in this column; however, each public library provided data for a 12-month period.

2October 2011 to September 2012.

3January 2011 to June 2012.

4January 2011 to December 2012.

5April 2011 to December 2012.

6December 2010 to September 2012.

7October 2010 to December 2012.

8July 2011 to December 2012.

9March 2011 to December 2012.

10February 2011 to December 2012.

11October 2010 to June 2012.

SOURCE: Institute of Museum and Library Services, Survey of Public Libraries in the United States, Fiscal Year 2012.

Survey Population Items

The PLS has three population items: (1) Population of Legal Service Area for each public library, (2) Total Unduplicated Population of Legal Service Areas for each state, and (3) State Total Population Estimate.

The population data are provided by the state library agency. The methods of calculation of the first two items vary significantly among states, and the state reporting periods also vary. The Total Unduplicated Population of Legal Service Areas does not include unserved areas and may vary from data provided by sources using standard methodology (e.g., the Census Bureau).

The total Population of Legal Service Area for all public libraries in a state may exceed the state’s Total Unduplicated Population of Legal Service Areas or the State Total Population Estimate. This happens in states where there are overlaps in population of legal service areas served by individual libraries, resulting in the same population being counted twice. Twenty-nine states had such overlapping service areas in FY 2012 (Table A-2).

Table A–2. States with public libraries with overlapping

Table B–2. service areas: Fiscal Year 2012

Arkansas Montana

Arizona New Hampshire

Colorado New Jersey

Connecticut New Mexico

Florida New York

Idaho North Dakota

Iowa Ohio

Indiana Oklahoma

Louisiana Rhode Island

Maine South Dakota

Maryland Texas

Massachusetts Vermont

Michigan Virginia

Minnesota Puerto Rico

Mississippi

SOURCE: Institute of Museum and Library Services, Survey of Public Libraries in the United States, Fiscal Year 2012.

To enable meaningful state comparisons using total Population of Legal Service Area data (for example, the number of print materials per capita), the Population of Legal Service Area data were adjusted to eliminate duplicative reporting due to overlapping service areas. The Public Library Data File includes a derived unduplicated population of legal service area figure for each library for this purpose (the variable POPU_UND). This value was prorated for each library by calculating the ratio of a library's Population of Legal Service Area to the state's total Population of Legal Service Area and applying the ratio to the state's Total Unduplicated Population of Legal Service Areas. (The latter item is a single, state-reported figure found on the Public Library State Summary/State Characteristics Data File. The variable name is the same on the state file: POPU_UND.)

Confidentiality

Two separate laws cover the protection of the confidentiality of individually identifiable information collected by the Institute of Museum and Library Services - the Privacy Act of 1974 and the E- Government Act of 2002. The Guidelines for Ensuring and Maximizing the Quality, Objectivity, Utility, and Integrity of Information Disseminated by the Institute of Museum and Library Services are prepared under the Treasury and General Government Appropriations Act for Fiscal Year 2001, Section 515(b). IMLS releases data to the public to use for statistical purposes only. Record matching or deductive disclosure by any user is prohibited. Procedures for disclosure avoidance were used in preparing public-use data files and tables of salary information for release.

Public- and Restricted-Use Data Files

The Public Library data file is available in two versions: public-use data and restricted-use data. Public-use data are available to all users. The public-use data file has some data suppressed to protect privacy and to prevent disclosure of personal information. If there is a need for access to suppressed information, data users should contact IMLS about procedures for obtaining access to the restricted-use data files.

Public-use data. On the public-use Public Library data file, selected expenditures data (i.e., Salaries, Employee Benefits, Total Staff Expenditures, and Other Operating Expenditures) of public libraries have been removed (i.e., the field is blank) when their total full-time equivalent (FTE) staff is less than or equal to 2.00, to protect the confidentiality of respondents. These data may also be suppressed for other libraries to ensure that all states that have suppressed data have a minimum of three suppressed records. The library’s Total Operating Expenditures and Other Expenditures Data are not affected by the suppression of these data. No data are suppressed on the public-use versions of the Public Library State Summary/State Characteristics Data File or Public Library Outlet Data File.

Restricted-use data. No data are suppressed on the restricted-use versions of the Public Library data file, Public Library State Summary/State Characteristics data file, or Public Library Outlet data file. The inclusion of all expenditures data, irrespective of the number of FTE staff, enables the identification of individual salary data for some libraries. Researchers requiring access to the restricted-use data must contact IMLS to obtain use of the data.

Data Collection

The FY 2012 PLS was released to the states over the Internet on December 19, 2012. States were placed into one of three reporting groups, based on their fiscal cycles or claim of extraordinary reporting hardship. States reported their data over the Internet via a web-based reporting system called WebPLUS (Web Public Library Universe System). WebPLUS was developed by the Census Bureau (the data collection agent). Edit follow-up was completed in late October of 2013. The editing process is described below.

Editing

State level. The respondent generates an Edit Report following direct data entry or import of their data into WebPLUS. The Edit Report, which can be viewed on-screen or printed, is used to identify and correct any errors, and to confirm the accuracy of data that generated edit warnings but required no change, before submitting the final file to the Census Bureau. In the FY 2012 PLS, four types of edit checks were performed:

1. Relational edit checks. This is a data consistency check between related data elements. For example, an edit message is generated if the number of “ALA-MLS” Librarians (librarians with master’s degrees from programs of library and information studies accredited by the American Library Association) is greater than “Total Librarians”.

2. Out-of-range edit checks. This is a range check that compares the data reported for an item to the

“acceptable range” of numeric values for the item. For example, an edit message is generated if average Public Service Hours per outlet per week is less than 11 or greater than 130, or if the current year/past year change in Children’s Circulation is less than 0.30 or greater than 3.44.

Survey due dates for groups 1, 2, and 3, were April 10, July 31, and August 21, 2013, respectively.

3. Arithmetic edit checks. This is an arithmetical accuracy check of a reported total and its parts to the generated total. For example, an edit message is generated if Total Operating Revenue is not equal to the sum of its parts (Local Government Revenue, State Government Revenue, Federal Government Revenue, and Other Revenue).

4. Blank, zero, or invalid data edit checks. This is a check of reported data against acceptable values.

For example, an edit message is generated if the Reporting Period Start Date is missing, or Print Materials is 0, or the Legal Basis Code is not a valid code.

The WebPLUS application generates state summary tables (showing state totals for all numeric data items) and single-library tables (showing data for individual public libraries in a state). State item response tables are also generated. Respondents were encouraged to review the tables for data quality issues before submitting their data to IMLS. State data submissions also included a signed form from the Chief Officer of the State Library Agency certifying the accuracy of the data.

National level. The Census Bureau and IMLS reviewed and edited the state data submissions, working closely with the PLS State Data Coordinators.

Imputation

Imputation is a procedure for estimating a value for a specific data item where the response is missing.

This section describes the imputation methods that were used to fill in the missing data items for the FY 2012 survey year.

The responding and non-responding libraries were sorted into imputation cells based on region codes developed by the Bureau of Economic Analysis and the size of the population. Each state was assigned a BEA region code (e.g. 01- New England (CT ME MA NH RI VT)). The cumulative root frequency method was used to determine the imputation cells. In survey year FY 2012, a total of 53 data items were eligible for imputation.

The imputation for non-responding libraries was performed using the data calculated from respondents in their imputation cells. Item imputation was performed on each record with nonresponse variables.

Following are descriptions of each imputation method used for the Public Libraries Survey (PLS).

Imputations were performed in two stages. In the first stage, imputations were carried out for nearly all missing values using the following methods: prior year times cell mean growth rate, adjusted cell mean, cell mean, prior year ratio, cell median ratio, direct substitution of prior year data, cell median, and special imputations. In the second stage, imputed values are adjusted for some missing values (based on the variable) using the following methods: obtained value by relationship of total to detail items, raking of detail items to total, special imputations, and consistency checks.

First Stage: Initial imputations

For each missing value, one of these methods of imputation was used to fill out the FY 2012 PLS dataset.

METHOD 1: Prior year value multiplied by cell mean growth rate. The imputed value for a missing item is calculated by the cell mean growth rate. The mean growth rate is calculated within a cell of libraries having similar characteristics using the average of all libraries with reported current year and prior year values for the missing item. In this method, prior year reported data can be as old as four years.

METHOD 2: Adjusted cell mean. The imputed value for a missing item is equal to the mean of all libraries reporting a current year value for that variable within an imputation cell multiplied by an

Cochran, W. (1977). Sampling Techniques, 3 rd

ed. Wiley. (p. 129) adjustment factor. The adjustment factor is the ratio of a library’s population of legal service area to the mean population of legal service area for libraries within the imputation cell.

METHOD 3: Cell mean. The imputed value for a missing item is equal to the mean of all libraries reporting a current year value for that variable within an imputation cell.

METHOD 4: Prior year ratio to another item. The imputed value for a missing item is equal to the library’s reported current year value of a highly correlated variable multiplied by a ratio. The ratio uses prior year reported values for that library of the missing item divided by the highly correlated variable. In this method, prior year reported data can be as old as four years.

METHOD 5: Current year cell median ratio to another item. The imputed value for a missing item is equal to the library’s reported current year value of a highly correlated variable multiplied by a cell median ratio value. The median ratio value is calculated for all libraries within an imputation cell having reported current year values for the variables in the ratio. Ratios are calculated using reported current year values for the missing item and the highly correlated variable.

METHOD 6: Direct substitution of prior year reported data. The imputed value for a missing item uses a library’s reported prior year data for that variable. In this method, reported prior year data can be as old as four years.

METHOD 7: Cell median. The imputed value for a missing item is equal to the median value of all libraries reporting a current year value for that variable within an imputation cell.

METHOD 8: Special impute for an item. The imputed value of an item is adjusted using its relationship with another reported item.

Each group of these PLS variables uses different methods of imputations:

Group One:

A group of PLS variables was imputed using method 1 (prior year multiplied by cell mean growth rate); if prior year reported data were not available then method 2 (the adjusted cell mean) was used; if the adjusted cell mean could not be calculated then method 3 (the cell mean) was used. Variables in this first group include:

Librarians with Master’s degrees from programs accredited by the ALA (MASTER)

Total Librarians (LIBRARIA)

All Other Paid Staff (OTHPAID)

Local Government Revenue (LOCGVT)

Total Operating Revenue (TOTINCM)

Salaries & Wages Expenditures (SALARIES)

Total Staff Expenditures (STAFFEXP)

Other Materials Expenditures (OTHMATEX)

Total Collection Expenditures (TOTEXPCO)

Other Operating Expenditures (OTHOPEXP)

Total Operating Expenditures (TOTOPEXP)

Print Materials (BKVOL)

Library Visits (VISITS)

Total Circulation (TOTCIR)

Total Attendance at Library Programs (TOTATTEN)

Number of Registered Borrowers (REGBOR)

Current Print Serial Subscriptions (SUBSCRIP)

Reference Transactions (REFERENC)

The Electronic Materials Expenditures (ELMATEXP) and Number of Internet Computers Used by General Public (GPTERMS) variables were both considered part of both groups one and six. If a missing value for ELMATEXP was imputed using method 4, then the Total Collection Expenditures (TOTEXPCO) variable was used as the other item in the prior year ratio. If a missing value for GPTERMS was imputed using method 4, then the number of Library Visits (VISITS) was used as the other item in the prior year ratio.

The Number of Children’s Programs (KIDPRO) and the Interlibrary Loans Received From (LOANFM) variables can both be considered part of both groups one and seven. If missing values for KIDPRO are imputed using method 5, then the Total Number of Library Programs (TOTPRO) variable is used as the highly correlated value in the cell median calculations. If missing values for LOANFM are imputed using method 5, then the Interlibrary Loans Provided To (LOANTO) variable is used as the highly correlated value in the cell median calculations.

Group Two:

A second group of variables was imputed using method 2 (adjusted cell mean); if the adjusted cell mean could not be calculated, then method 3 (the cell mean) was used. These variables include:

Total Capital Expenditures (CAPITAL)

Local/Other Databases (DB_LO_OT)

Group Three:

A third group of variables was imputed using method 1 (prior year multiplied by cell mean growth rate); if prior year reported data were not available to calculate growth rates, then method 7 (cell median) was used. These variables include:

State Government Revenue (STGVT)

Other Operating Revenue (OTHINCM)

Local Government Capital Revenue (LCAP_REV)

Total Capital Revenue (CAP_REV)

Electronic Books (EBOOK)

Group Four:

A fourth group of variables was imputed using method 6 (direct substitution of prior year reported data.

These variables include:

Databases (DATABASE)

Number of Young Adult Programs (YAPRO)

Young Adult Program Attendance (YAATTEN)

If prior year data were not available for DATABASE, then method 3 (the cell mean) was used. If prior year data were not available for YAPRO or YAATTEN, then method 2 (adjusted cell mean) was used; if the adjusted cell mean could not be calculated, then method 3 (cell mean) was used.

Group Five:

A fifth group of variables was imputed using only method 7 (cell median); these variables were:

Federal Government Revenue (FEDGVT)

State Government Capital Revenue (SCAP_REV)

Federal Government Capital Revenue (FCAP_REV)

Other Capital Revenue (OCAP_REV)

Group Six:

A sixth group of variables was imputed using method 4 (prior year ratio to another item); if nonzero prior year reported unit data were not available, then method 2 (adjusted cell mean) was used; if the adjusted cell mean could not be calculated, then method 3 (cell mean) was used. The variables in the sixth group, along with the highly correlated variable(s) in the prior year ratio, include:

Employee Benefits Expenditures (BENEFIT):

o Total Staff Expenditures (STAFFEXP)

Inter-Library Loans Provided To (LOANTO):

o Inter-Library Loans Received From (LOANFM)

The Electronic Materials Expenditures (ELMATEXP) and Number of Internet Computers Used by General Public (GPTERMS) variables were both considered part of both groups one and six. If a missing value for ELMATEXP was imputed using method 4, then the Total Collection Expenditures (TOTEXPCO) variable was used as the other item in the prior year ratio. If a missing value for GPTERMS was imputed using method 4, then the number of Library Visits (VISITS) was used as the other item in the prior year ratio.

The Inter-Library Loans Provided To (LOANTO) can be considered part of both groups six and seven.

If nonzero prior year reported unit data were available, then method 4 (prior year ratio to another item) was used; if not, then method 5 (cell median ratio with another item) was used as the backup method.

For both methods, the Interlibrary Loans Received From (LOANFM) was used as the highly correlated variable when forming the ratios.

The Public Service Hours per Year (HRS_OPEN) variable is similar to the other variables in group six, and used method 4 (prior year ratio to another item) to impute for missing values. If a missing value for HRS_OPEN was imputed using method 4, then the correlated variable in the prior year ratio was an internal variable (TOTOUT), that was the sum of the following three variables: Number of Branch Libraries (BRANLIB), Number of Bookmobiles (BKMOB) and Number of Central Libraries (CENTLIB). If the missing value for HRS_OPEN could not be imputed using method 4, then method 8 (special imputation) was used instead. To impute for missing HRS_OPEN variables using special imputations, the internal TOTOUT variable was multiplied by the mean HRS_OPEN value divided by the sum of three items: the mean CENTLIB, mean BRANLIB and mean BKMOB values.

Group Seven:

A seventh group of variables was imputed using method 5 (cell median ratio with another item); if current year reported unit data were not available for the other variable in the ratio, then method 2 (adjusted cell mean) was used ; if the adjusted cell mean could not be calculated, then method 3 (cell mean) was used.

The variables in the seventh group, along with the highly correlated variable(s) used in the cell median ratio, include:

Print Material Expenditures (PRMATEXP):

o Total Collection Expenditures (TOTEXPCO)

State Databases (DB_ST):

o Total Licensed Databases (DATABASE)

Circulation of Children’s Materials (KIDCIRCL):

o Total Circulation (TOTCIR)

Total Number of Library Programs (TOTPRO):

o Number of Children’s Programs (KIDPRO)

Children’s Program Attendance (KIDATTEN):

o Total Attendance at Library Programs (TOTATTEN)

Number of Users of Public Internet Computers Per Year (PITUSR):

o Number of Internet Computers Used by General Public (GPTERMS)

The Number of Children’s Programs (KIDPRO) and the Interlibrary Loans Received From (LOANFM) variables can both be considered part of both groups one and seven. If missing values for KIDPRO are imputed using method 5, then the Total Number of Library Programs (TOTPRO) variable is used as the highly correlated value in the cell median calculations. If missing values for LOANFM are imputed using method 5, then the Interlibrary Loans Provided To (LOANTO) variable is used as the highly correlated value in the cell median calculations.

The Inter-Library Loans Provided To (LOANTO) can be considered part of both groups six and seven.

If nonzero prior year reported unit data were available, then method 4 (prior year ratio to another item) was used; if not, then method 5 (cell median ratio with another item) was used as the backup method.

For both methods, the Interlibrary Loans Received From (LOANFM) was used as the highly correlated variable when forming the ratios.

Group Eight:

An eighth group of variables was introduced in FY 2010.

Audio – Physical Units (AUDIO_PH) Audio – Downloadable Titles (AUDIO_DL) Video – Physical Units (VIDEO_PH) Video – Downloadable Titles (VIDEO_DL)

The Audio – Physical Units (AUDIO_PH) and Audio – Downloadable Titles (AUDIO_DL) variables are both subsets of the former PLS AUDIO variable. Similarly, the Video – Physical Units (VIDEO_PH) and Video – Downloadable Titles (VIDEO_DL) variables are both subsets of the former PLS VIDEO variable. Imputations for variables in group eight were done using method 8 (Special imputes), which depended on prior year reported values. For FY 2011 and FY 2010, these reported prior values used the AUDIO_PH and VIDEO_PH variables. For FY 2009 and FY 2008 , the former AUDIO and VIDEO variables were used as reported prior values. If prior year data were not available, then method 2 (adjusted cell mean) was used to carry out the imputations. If the adjusted cell mean could not be calculated, then method 3 (cell mean) was used.

Second Stage: Adjustments

The methods listed below adjusted imputations for some of the missing values, based on the PLS variable. These adjustments verify that detail items sum to totals, making corrections as necessary.

METHOD 9: Obtained value by relationship of total to detail items.

The imputed value of a total was adjusted using its relationship with reported detail items.

METHOD 10: Raking of detail items to match total.

The imputed value for a detail item was adjusted by raking methods so that it matches a reported total.

METHOD 11: Changed by consistency check. The imputed value was adjusted using customized consistency checks specific to that variable.

Each group of these PLS variables uses different methods to adjust imputations:

Group One:

Imputed values for these variables were checked for anomalies and adjusted when necessary using methods 9 (obtained value by relationship of total to detail items) or method 10 (raking) after the initial stage of imputation:

All Other Paid Staff (OTHPAID)

Local Government Capital Revenue (LCAP_REV)

State Government Capital Revenue (SCAP_REV)

Federal Capital Revenue (FCAP_REV)

Other Capital Revenue (OCAP_REV)

Group Two:

Imputed values for these variables were checked for anomalies and adjusted when necessary using method 9 (obtained value by relationship of total to detail items) after the initial stage of imputation:

Salaries & Wages Expenditures (SALARIES)

Employee Benefits Expenditures (BENEFIT)

Total Staff Expenditures (STAFFEXP)

Total Paid Employees (TOTSTAFF)

Print Materials Expenditures (PRMATEXP)

Total Collection Expenditures (TOTEXPCO)

Other Operating Expenditures (OTHOPEXP)

Total Operating Expenditures (TOTOPEXP)

Total Capital Revenue (CAP_REV)

Total Operating Income (TOTINCM)

Total Licensed Databases (DATABASE)

Note that in certain sparse data conditions, imputed values for Salaries & Wages Expenditures (SALARIES), Employee Benefits Expenditures (BENEFIT), Total Staff Expenditures (STAFFEXP), and Total Collection Expenditures (TOTEXPCO) were adjusted when necessary using a prior year ratio to another item (method 4) after the initial stage of imputation.

Group Three:

method 10 (raking) after the initial stage of imputation:

Print Materials Expenditures (PRMATEXP)

Electronic Materials Expenditures (ELMATEXP)

Other Materials Expenditures (OTHMATEXP)

Federal Government Revenue (FEDGVT)

State Government Revenue (STGVT)

Local Government Revenue (LOCGVT)

Other Operating Revenue (OTHINCM)

State Databases (DB_ST)

Local/Other Databases (DB_LO_OT)

Group Four:

method 11 (consistency check) after the initial stage of imputation:

Total Circulation (TOTCIR)

Total Number of Library Programs (TOTPRO)

Number of Children’s Programs (KIDPRO)

Number of Young Adult Programs (YAPRO)

Total Attendance at Library Programs (TOTATTEN)

Children’s Program Attendance (KIDATTEN)

Young Adult Program Attendance (YAATTEN)

Number of Internet Computers Used by General Public (GPTERMS)

Number of Users of Public Internet Computers Per Year (PITUSR)

Imputed values for Total Capital Expenditures (CAPITAL) were checked for anomalies and adjusted when necessary using method 8 (special imputes) after the initial stage of imputation.

Imputed values for Total Librarians (LIBRARIA) were checked for anomalies and adjusted when necessary using methods 9 (obtained value by relationship of total to detail items), 10 (raking) or 11 (consistency check) after the initial stage of imputation.

Imputed values for Librarians with Master’s degrees from programs accredited by the ALA (MASTER) were checked for anomalies and adjusted when necessary using methods 9 (obtained value by relationship of total to detail items) or 11 (consistency check) after the initial stage of imputation.

Note that imputed values for Total Number of Library Programs (TOTPRO), Children’s Program Attendance (KIDATTEN), Circulation of Children’s Materials (KIDCIRCL) and Interlibrary Loans Provided To (LOANTO) were checked for anomalies and adjusted when necessary using cell median ratio to another item (method 5).

Non-sampling Errors

Because all units in the universe are surveyed, the data are not subject to sampling error, but they are subject to non-sampling errors, such as errors in response, non-response errors, coverage errors arising from an incomplete listing of public libraries, coding errors, or processing errors.

Every effort is made to mitigate such errors. The editing efforts described above are designed to decrease the number of errors due to inaccurate response or due to processing problems. Imputation lessens the effect of non-response. Efforts are made to obtain complete listings of public libraries from the state library agencies. Although such efforts are made, some non-sampling error likely remains in the data.

Note: Errors in response to the audio and video downloadable data were confirmed by some states. The data were incorrectly reported as ‘units’ instead of ‘titles’. The incorrect data for these states were deleted from the data files.

Geocoding

As part of the post-processing of the data files, supplemental geographic information is provided for each record, where possible. This supplemental geographic information is available for administrative entities and outlets. All supplemental geography codes are assigned based on the address information reported by the respondent. The following supplemental geography information is provided with the release of the PLS FY 2012 data:

Longitude Latitude FIPS State Code FIPS County Code FIPS Place Code County Population Locale Code Census Tract Census Block Congressional District Core based statistical area Metropolitan/Micropolitan area flag Geocoding Accuracy Level (GAL) GAL Match Status Postal Match Status

The geocodes were processed in batch using an online geocoding service. Physical addresses for administrative entities and outlets (state, city, address, zip code, zip code + 4) were input into the server, processed and sent back to IMLS as a text file.

There are six different geographic levels at which and address could be matched: address point, house, intersection, extended postal code (ZIP+4), street, and postal code (ZIP). This denotes the accuracy of the matched results e.g. geocoding accuracy levels (GALs). An address point match means that the input address matched to an exact latitude/longitude location. A house match means that the input address matched to a house number along an address range. An intersection match means that the input address matched to two streets that intersect. An extended postal code (postalcode-sub) match means that the input address matched to an extended postcode (ZIP+4). A street match means that the input address matched to a street edge in the database i.e. the midpoint of a street chain within a city. A postal code match (postalcode-main) means that the input address matched to a main postal code (ZIP). When the geocoding service failed to match, records were processed manually. These records were identified as

ALT.

Of the 9,305 libraries in the 50 states, the District of Columbia, Puerto Rico and Guam that appear in the administrative entity level, 33.7 percent of the records were address point matches, 56.5 percent were house matches, 0.2 percent were intersection matches, 5.0 percent were extended postal code matches,

2.1 percent were street matches, and 2.6 percent were postal code matches. Only 2 records failed to match and were manually located.

Of the 17,586 central, branch, bookmobiles, and books by mail only outlets in the 50 states, the District of Columbia, Puerto Rico and Guam that appear in the outlet-level file, 38.2 percent were address point matches, 51.1 percent were house matches, 0.1 percent were intersection matches, 5.2 percent were extended postal code matches, 2.4 percent were street matches, and 2.9 percent were postal code matches. Only 9 records failed to match and were manually located.

Locale codes are added to the data files as part of post-processing (first added to the PLS files in FY2008). Locale codes allow users to identify whether or not library outlets and administrative entities are located in cities, suburbs, towns, or rural areas. These codes were assigned to libraries using the same methodology that is used to assign public schools locale codes in the National Center for Education Statistics’ Common Core of Data datasets. This locale coding system classifies areas into four major types – city, suburban, town, and rural – each with three subcategories. The 12 different locale codes and the criteria for their assignment are as follows:

11 - City, Large: Territory inside an urbanized area and inside a principal city with population of

250,000 or more.

12 - City, Midsize: Territory inside an urbanized area and inside a principal city with population less than 250,000 and greater than or equal to 100,000.

13 - City, Small: Territory inside an urbanized area and inside a principal city with population less than 100,000.

21 - Suburb, Large: Territory outside a principal city and inside an urbanized area with population of 250,000 or more.

22 - Suburb, Midsize: Territory outside a principal city and inside an urbanized area with population less than 250,000 and greater than or equal to 100,000.

23 - Suburb, Small: Territory outside a principal city and inside an urbanized area with population less than 100,000.

31 - Town, Fringe: Territory inside an urban cluster that is less than or equal to 10 miles from an urbanized area.

32 - Town, Distant: Territory inside an urban cluster that is more than 10 miles and less than or equal to 35 miles from an urbanized area.

33 - Town, Remote: Territory inside an urban cluster that is more than 35 miles from an urbanized area.

41 - Rural, Fringe: Census-defined rural territory that is less than or equal to 5 miles from an urbanized area, as well as rural territory that is less than or equal to 2.5 miles from an urban cluster.

42 - Rural, Distant: Census-defined rural territory that is more than 5 miles but less than or equal to 25 miles from an urbanized area, as well as rural territory that is more than 2.5 miles but less than or equal to 10 miles from an urban cluster.

43 - Rural, Remote: Census-defined rural territory that is more than 25 miles from an urbanized area and is also more than 10 miles from an urban cluster.

Library outlets were assigned locale codes based on the geocoded latitude and longitude values of their street addresses. Administrative entities were assigned locale codes based on the modal locale codes among central and branch libraries of that library system (excluding bookmobiles and books-by-mail).

Whenever there was a tie, the administrative entity retained its prior local code, if that code was among the tied values. If the tie involved locale codes that were different from the locale code corresponding to the administrative entity’s prior locale code, the most urban code of the tied locale codes was assigned to that administrative entity. The order of urbanicity of locale codes is from the highest, most urban (11) to the lowest, most rural (43).

The geocoding service returned the following geographic identifiers that appear on the administrative entity and outlet-level files: longitude, latitude, FIPS state code, FIPS county code, FIPS place code, Census tract, Census block, core based statistical area, metropolitan/micropolitan area flag, geocoding accuracy level (GAL), GAL match status, and postal match status. The Census Bureau added locale codes, Congressional district codes, and county population estimates to the file. The World Geodetic System 1984 (WGS 1984, Mercator Web) map projection was used throughout the geocoding process.

II. B. Guidelines for Processing the Data Files

See Introduction, Confidentiality, and Public- and Restricted-Use Data Files above for a description of the files.

The Census Bureau developed the software that the states used to report the PLS data. At survey mail-out, all numeric data fields were initialized with “-2”, and respondents were instructed to replace the “-2” with valid data. Alphanumeric fields that are blank or that contain “M” and numeric fields that contain “-1” indicate nonresponse. A zero (0) response is reported data and indicates the library or outlet had none of the item.

For the 50 states and the District of Columbia, missing data for numeric items were imputed.

Missing data were not imputed for outlying areas. See the Imputation section above for a discussion of the imputation methodology. See Appendix G for imputation flags and their definitions.

How to remove imputed values from the data.