Supporting Statement B

B. Collections of Information Employing Statistical Methods

1. Respondent Universe and Sampling Methods

At each of the five Demonstration sites, the survey will be administered to up to 100 participants and 80 non-participant eligibles who were offered but refused participation (up to 900 total sample). Given the current enrollment pace of the sites, and the time remaining in the Demonstration, we will need to attempt to survey all participants to achieve the sample size required. Since most beneficiaries offered the option to participate turn down the invitation at most sites, we may sample decliners if the number of decliners is sufficiently large. Selection criteria for the beneficiary sample include: 1) use of adult day care services in the two weeks prior to joining the Demonstration or declining to join, and 2) gender.

2. Procedures for the Collection of Information

The general approach to collecting data through the survey was described in the evaluation Design Report to CMS (Appendix A).

3. Methods to Maximize Response Rates and Deal with Issues of Non-response

Issues of non-response to the survey are minimized by CMS’ provision of accurate contact information for respondents. In addition, the introductory letter from the CMS Privacy Board will be useful in encouraging participation. During the phone calls for the case study interviews, it was found that most respondents remembered the letter. Also, stating that we were from a university and that we were working for Medicare to help improve the Medicare program helped to overcome skepticism.

Based on prior experience conducting telephone surveys of elders who were associated with service programs, we expect a positive response rate of 75% among the beneficiaries we invite to participate. We will continue to invite participation until the total sample of 90 beneficiaries per Demonstration site is achieved.

4. Tests of Procedures or Methods to be Undertaken

The surveys were pre-tested on small samples of participants (4) and decliners (4), which led to modifications in the versions tested. The final versions consist of 24 questions for participants and 14 questions for decliners. None of the questions appeared sensitive to respondents. The time to administer the test version for participants and their proxies ranged from 7 to 15 minutes. The time to administer the test version for decliners and their proxies ranged from 4 to 6 minutes. We expect the final version of the participant survey to be at least a minute shorter since we removed two multi-item scaled questions in favor of short, open-ended questions.

We will use appropriate multivariate techniques to analyze the factors associated with satisfaction and out-of-pocket costs, as well as differences in satisfaction and costs between participants and decliners. Independent variables will include those found in Table 1 data (site, gender, prior use of adult day care) and those from the survey itself (demographics, living situation, needs, helpers, race/ethnicity, age, and Medicaid status).

5. Statistical Contact

Grant Ritter, PhD

Senior Scientist

Brandeis University

Schneider Institute for Health Policy

415. South Street, Waltham, MA 02454

e-mail address - [email protected]

telephone number (781) 736-3872.