The goal of the sampling design was to select sufficient numbers of organizations to complete 200 telephone interviews, while maximizing the response rate. A sample of this size is large enough to investigate what services were delivered after hurricanes Katrina and Rita and the collaborations used.
The telephone survey was conducted in three geographic areas: the states of Louisiana and Mississippi and an area 50 miles in radius around the Houston Astrodome where many evacuees from New Orleans were sheltered.
Because there are no existing lists of FBCOs in the region or nationally, we used two independent sources to identify organizations of interest and create a master list from which to draw the sample. The American Church List (ACL), a database of religious congregations, is regarded as the most complete and current information on congregations in the United States. The ACL contains contact information, such as address and telephone number for congregations, and, for some entries, demographic information about the congregation. The list, purchased in July 2007, had 14,213 religious organizations in Louisiana, Mississippi, and a 50-mile radius around the Houston Astrodome. The National Center for Charitable Statistics (NCCS) database contains information on 501(c)(3) organizations that have revenues of $25,000 or more and file annual Forms 990 with the U.S. Internal Revenue Service. The latest NCCS data available (2005) identified 2,957 human service nonprofits in the geographic areas of study.
Two considerations shaped the sample design. First, it was assumed that the intensity of the storms impact affected the ability of FBCOs to provide relief and recovery services. FBCOs closest to the heavily damaged areas might be less able to respond than those farther away. To adjust for this likelihood, the sample was divided into three strata: (1) areas directly impacted by the storm, (2) those partially impacted or adjacent to directly impacted areas, and (3) the remainder of the state. All FBCOs selected from Houston were regarded as part of the third (or distant) stratum.
Second, prior research suggested that congregations would be more difficult to reach by telephone than nonprofit groups. So, the sampling plan called for oversampling congregations to achieve an adequate number of them in the final sample.
Proportional random samples of congregations and nonprofits, stratified by state and impact area, were drawn from the master list derived from the ACL and NCCS databases. This selection process produced a sample of 615 organizations. This pool of potential respondents contained both large and small congregations and nonprofits.