Once the survey was developed, the population was taken from the information system database. The Survey Research Center identified clinical managers who also were supervisors by comparing the names of the clinical managers with the names of the supervisors. Each supervisor in the study received a packet that included: (1) a cover letter explaining the survey and providing the information for informed consent (Appendix C); (2) a survey (Appendix D); (3) a $2 cash incentive; and (4) a postage paid business reply envelope addressed to the Survey Research Center. We considered a higher incentive but decided to offer the same amount as that given for the Direct Care Worker Survey, so as not to adversely affect the DCW response rate. Prior research has shown that incentives given upfront to all potential participants are more effective in increasing response rates than incentives paid later only to respondents (Church, 1993; James and Bolstein, 1992).
The Survey Research Center alerted its contact at each provider before sending survey packets for distribution. The contact chose one of three ways to distribute the surveys -- at staff meetings, with paychecks, or in worker mailboxes at the organization.
The cover letter informed respondents that their participation was voluntary and that their responses would be kept confidential. Respondents provided passive consent by completing and returning the survey. Respondents returned completed surveys directly to the Survey Research Center to ensure that respondents employers would not see their responses. The Survey Research Center tracked survey responses using identification numbers on each survey.
Prior research has shown that multiple mailings increase response rates (Dillman, 2000). Therefore, approximately a month after the first survey administration, follow-up packets were sent to each organization. Supervisors who had not completed a survey received a packet containing a follow-up cover letter (Appendix E), another copy of the survey, and another business reply envelope. To ensure that employers could not distinguish non-respondents from previous respondents, packets also were sent to supervisors who had completed the survey. These packets contained a cover letter expressing our thanks for completing the survey, along with a copy of the BJBC newsletter.
As the surveys were returned, they were logged and scanned into the system using the identification number printed on the surveys. At the end of the fielding period, the Survey Research Center verified and cleaned the data and removed all identifying information before providing the data files to the research team.
The original Supervisor Survey project was designed to be cross-sectional, with a one-time administration scheduled toward the end of the BJBC demonstration. After approval of this project, the Robert Wood Johnson Foundation and the Atlantic Philanthropies, the funding agencies for the evaluation of the BJBC demonstration, authorized the Penn State Survey Research Center to use funding from the evaluation for a baseline (Time 1) administration of the survey. However, at this point in the project it was too late for four of the states to receive a baseline survey. Therefore, the Time 1 administration was conducted only in North Carolina, which had experienced delays in starting the demonstration.
The Time 1 administration in North Carolina occurred from November 2005 through March 2006. The second administration in North Carolina and the originally-planned administration in the other four BJBC states (Time 2) took place from July 2006 through June 2007. Because of the larger sample size, most analyses were conducted using the cross-sectional data from the Time 2 administration. However, when appropriate, analyses employed the baseline data from North Carolina, including a small panel of supervisors who responded to the survey in both time periods.