Publication Type: Conference Paper/Unpublished Manuscript
Review Method: Peer Reviewed
Abstract: Efficiently discovering rare populations of interest is an obstacle often encountered in survey research. This paper examines disparate approaches used by the CDC-sponsored REACH 2010 Risk Factor Survey to interview rare populations in two urban settings. In the first instance an area probability methodology was utilized to interview Cambodian households in Lowell, Massachusetts in 2002. Cambodians comprised a very small portion of the city’s population but even less so of the total households, which had a significant impact on data collection costs. The second round of the survey used a stratified telephone sampling procedure which will be compared to the field method.
In the second instance a GIS (Geographic Information Systems) – based telephone sampling approach was utilized to identify African-American and Hispanic households in Hillsborough County, New Hampshire. These two populations together comprised 4 percent of the total county population. Using a sophisticated sampling method, we were able to realize a racial eligibility rate of 11 percent among sample households which allowed telephone interviewing to be conducted instead of a more costly field interviewing approach. This paper examines the cost and effectiveness of each approach, explores possibilities of other methods (such as adaptive sampling), and discusses their applicability to discovering rare populations in similar urban settings.