Survey Methodology and Data Reliability

In late March 1989, questionnaires (see attachment) were mailed to a national probability sample of 846 public libraries from a universe of approximately 8,500 main libraries and 5,600 branch libraries. The sample included 345 main libraries without branches, 195 main libraries with branches, and 306 branch libraries. The data were collected for individual library buildings rather than for library systems. State libraries and cooperative systems were excluded from the survey. Telephone followup of nonrespondents was initiated in mid-April; data collection was completed in May with a response rate of 97 percent. The sampling frame used for the survey was the universe file of U.S. public libraries purchased from Market Data Retrieval. During data collection, it was discovered that some branch libraries were not represented on the frame, indicating some undercoverage in the Market Data Retrieval frame. The extent of undercoverage is not known, since this survey was not designed to estimate undercoverage.

The sample was allocated proportionally to main libraries and branches. An equiprobability sample of main libraries was drawn from each size of population stratum (less than 10,000; 10,000- 99,999; 100,000 or more). Then, Keyfitz procedures¹⁵ were used to draw the sample of branches from each size stratum to ensure that the sample of branch libraries overlapped minimally with branch libraries in library systems represented by main libraries selected into the sample. This procedure maximized the number of library systems represented in the sample. This is the same sample that was used in FRSS #28, Services and Resources for Young Adults in Public Libraries, released by NCES in July 1988. The survey data were weighted to reflect these sampling rates (probability of selection) and were adjusted for nonresponse. Numbers in the tables and text have been rounded. Percentages and averages have been calculated based on the actual estimates rather than the rounded values.

The standard error is a measure of the variability due to sampling when estimating a statistic. It indicates how much variance there is in the population of possible estimates of a parameter for a given size sample. Standard errors can be used as a measure of the precision expected from a particular sample. If all possible samples were surveyed under similar conditions, intervals of 1.96 standard errors below to 1.96 standard errors above a particular statistic would include the true population parameter being estimated in about 95 percent of the samples. This is a 95 percent confidence interval. For example, for the percentage of public libraries with the assistance of a children's coordinator or consultant available, the estimate for all libraries is 67.5 and the standard error is 1.4. The 95 percent confidence interval for this statistic extends from 67.5- (1.4 times 1.96) to 67.5 + (1.4 times 1.96) or from 64.8 to 70.2.

Estimates of standard errors were computed using a balanced half sampling technique known as balanced repeated replications. Estimated standard errors for some key variables are included in Table 8: Selected standard errors by library characteristic, Table 8 Continued, Table 8 Continued 2, Table 8 Continued 3. Standard errors for statistics not included in this table can be obtained upon request.

Relationships between variables with 2 or more levels have been tested using chi-square tests at the .05 level of significance, adjusted for average design effect. If the overall chi-square test was significant, it was followed up with tests using a Bonferroni t statistic, which maintained an overall 95 percent confidence level or better.

Some of the variables used to classify libraries were correlated (such as number of library users per week and library type). However, the sample size of this survey limits our ability to understand the full multivariate nature of the responses by correlated classification variables.

Survey estimates are also subject to errors of reporting and errors made in the collection of the data. These errors, called nonsampling errors, can sometimes bias the data. While general sampling theory can be used to determine how to estimate the sampling variability of a statistic, nonsampling errors are not easy to measure and usually require that an experiment be conducted as part of the data collection procedures or the use of data external to the study.

Nonsampling errors may include such things as differences in the respondents" interpretation of the meaning of the questions, differences related to the particular time the survey was conducted, or errors in data preparation. During the design of the survey and survey pretest, an effort was made to check for consistency of interpretation of questions and to eliminate ambiguous items. The questionnaire was pretested with respondents like those who completed the survey, and the questionnaire and instructions were extensively reviewed by NCES and by a panel of librarians with specialties in children's services. Manual and machine editing of the questionnaires was conducted to check the data for accuracy and consistency. Cases with missing or inconsistent items were noted and respondents recontacted by telephone; data were keyed with 100 percent verification.

Data are presented for all libraries and by the following library characteristics: number of library users per typical week, library type, availability of a children's librarian, number of hours open to the public per week, percentage of the library's book budget used for children's books, and the percentage of the library's total circulation that is children's materials. Classifications for number of library users per typical week are: light patronage--less than 200 patrons; moderate patronage--200-999 patrons; and heavy patronage--1000 or more patrons. Type classifications are as follows: main libraries without branches are those libraries which represent a single-library system; main libraries with branches are lending libraries which serve as system headquarters for a multilibrary system or that are located at the same address as the administrative office of an all-branch system (where no one library in the system has been designated as headquarters); and branch libraries are those libraries belonging to, but not headquarters for, a multi-library system.

Classifications for number of hours open to the public per week are: 1-30; 31-50; and 51-84. The classifications used for percent of the budget used for children's books are: O-25 percent; 26-39 percent; and 40-100 percent. Classifications for percent of total circulation that is children's materials are: O-33 percent; 34-49 percent; and 50- 100 percent. Data regarding number of library patrons per week, whether or not the library had a children's librarian, number of hours open to the public, percentage of the book budget used for children's books, and the percentage of the total circulation that is children's materials were obtained from the survey.

The survey was performed under contract with Westat, Inc., using the Fast Response Survey System (FRSS). Westat's Project Director was Elizabeth Farris, and the Survey Manager was Laurie Lewis. Jeffrey Williams was the NCES Project Officer. The data requester, who participated in the survey design and analyses, was Ray Fry, Senior Advisor for Library Programs, OERI. FRSS was designed to collect quickly, and with minimal burden on respondents, small quantities of data needed for education planning and policy.

The following consultants assisted with the planning of the survey: Jane Botham, Children's Services Coordinator, Milwaukee Public Library; Barbara Immroth, Associate Professor, Graduate School of Library and Information Science, University of Texas at Austin; Margaret Kimmel, Professor, School of Library and Information Science, University of Pittsburgh; Mary Jo Lynch, Director of the Office of Research, American Library Association; Susan Roman, Executive Director of the Association for Library Service to Children, American Library Association; and Mary Somerville, Coordinator of Children's Services, Broward County (Florida) Library.

The reviewers were Macknight Black, Mary Frase, Ray Fry, and Iris Silverman, U.S. Department of Education, and Elizabeth Hoke, former coordinator of children's services at Montgomery County (Maryland) public library.

For information about this survey or the Fast Response Survey System, contact Jeffrey Williams, Office of Educational Research and Improvement, National Center for Education Statistics, 555 New Jersey Avenue, NW, Washington, DC 20208-5651, telephone (202) 357-6333.

---

¹⁵For information about the Keyfritz procedure, see the following articles J. Michael Brick, David Morganstein, and Charles Wolters, "Additional Uses for Keyfitz Selection," Proceedings Of the Section on Survey Research Methods of the American Statistical Association, (1987): 787-791; and Nathan Keyfitz, "Sampling with Probabilities Proportionate to Size: Adjustment for Changes in Probabilities," Journal of the American Statistical Association, vol. 46 (1951): 105-109.

Top