Supporting Statement B.BR Audio Guide

Supporting Statement—Section B

1. Collection of Information Employing Statistical Methods

B.1. Respondent Universe

The sample of respondent universe for the evaluation study of the Audio Guides consists of multiple levels or populations. The respondent universe for the initial survey consists of all 16,500 libraries that receive the Audio Guide materials. We estimate approximately a 75% response rate from this initial survey for at total estimated sample of approximately 12,375 responses.

Second, we will be conducting an online survey with designated representatives of the libraries who returned the initial survey. The respondent universe for this online follow-up survey consists a sample of the libraries that complete the initial survey with email contact information. We will randomly sample 2,000 respondents from the returned initial survey pool using stratified random sampling procedures. The stratification criteria we will use will include library size, geographic region, type of library, and target audience served.  We anticipate a 75%% response rate for the follow-up interviews.

The respondent universe for the follow-up telephone interviews includes non-respondents from the initial survey and non-respondents from the follow-up survey. We anticipate interviewing approximately 90 in the first group and 30 in the second. Our interview samples will be stratified by geographical location, library size, and targeted library audience. We anticipate a 75%% response rate for the follow-up interviews.

B.2. Statistical Methodology

a. Stratification and Sample Selection

We will, as described above, create a sample for the online survey from the libraries that completed the initial survey with email contact information. We also plan to use sampling for the follow-up telephone interviews of non-respondents to both surveys and respondents to the online survey. Our sampling procedures will include purposive sampling to ensure maximum representation across the response universe.

For analysis purposes, we will sample open-ended responses on the initial and follow-up surveys since the analysis of large numbers of open-ended data are time and cost-prohibitive. We will sample every 10th response and create codes for the sampled responses. To ensure representativeness of the sample to the pool of responses, we will compare the characteristics of the libraries whose responses were sampled to those of the pool as a whole. Oversampling will be done to fill in areas of non-representativeness.

2. Estimation Procedures

The primary method of data analysis for the initial and follow-up online surveys and for quantitative indices on the telephone interviews will be descriptive. We will summarize responses by using means, standard deviations, and frequencies of responses. In some cases, we may conduct simple cross-tabs, mean comparisons, or correlations to assess relationships among particular items.

2. Degree of Accuracy

We anticipate that the degree of accuracy will be sufficient to meet the purposes described in the “Needs and Uses of the Data” section of the supporting statement.

2. Specialized Sampling

Not applicable

2. Less Frequent Data Collection

Not applicable

B.3. Response Rates

We will employ a variety of methods to maximize response rates to the initial and online follow-up surveys. For the initial survey, we will send out periodic reminders to libraries to complete and return the survey. These reminders may come from the evaluation firm, IMLS, or from emails or mailings from professional library organizations. For the online follow-up surveys, we will rely on email reminders to the libraries in the online survey sample. For the telephone interviews, we will minimize non-response with email and telephone reminders to respondents in the samples one week after the initial contact to schedule the interview.

B.4. Tests of Procedures

We plan to pilot all survey items and interview questions to ensure clarity, reduce redundancy, and assess the level of response burden on the respondents. We typically pilot surveys with 3 to 5 individuals. Online survey links are also piloted to ensure that there are no technological glitches in survey administration.

B.5. Contact Information

Contact information for the firm conducting the statistical analyses is:

Terri Akey, Ph.D.

Rockman et al

3925 Hagan St., Suite 301

Bloomington, IN 47401

812.333.8883

terri@rockman.com