Methods for Robust Data Collection on Small Populations: Case of the Pacific Islander Health Study
*Sela V. Panapasa, University of Michigan
Keywords: sample design, survey research methods, Pacific Islanders, Samoans, Tongans, faith-based organizations
National surveys play a critical role in the development of policy and evidence-based interventions. Unfortunately, nationally representative sampling methods fail to capture rare populations such as Pacific Islanders (PIs) in numbers sufficient to provide reliable population estimates. PIs represent over 25 distinct groups, cultures, languages and histories that make it even more difficult to collect robust information by subpopulation. To address these data limitations, this paper presents a viable and effective methodology for collecting robust data on two hard-to-reach PI subpopulations (Samoans and Tongans) as part of the Pacific Islander Health Study (PIHS). This paper will discuss methods for sampling design, data collection strategies that minimize data bias and error, and increase response rates. Using the Community-based Participatory Research Approach, respondents were derived from 20 multi-denominational registries for San Mateo and Los Angeles counties in California to construct a representative sample of Pacific Islander households. Faith-based organizations represented a promising strategy for identifying Samoan and Tongan households. Samoan and Tongan bi-lingual interviewers were trained to screen households and administer separate questionnaires to a randomly selected adult and adolescent respondent. The questionnaires, introductory letters and outreach materials were translated in Samoan and Tongan for effective communication with respondents. The paper argues that the use of innovative methods in the PIHS can obtain reliable and generalizable data on hard-to-reach populations. The approach also demonstrates the real application of survey methods and principles in the collection of robust data on Pacific Islanders.