Study Design and Sample
A cross-sectional survey of a random sample of adults residing in the four counties of Humboldt, Del Norte, Trinity and Mendocino was conducted. A four-page self-administered survey was developed by project administrators at CCRP. The survey contained questions about general health, mental health, preventive health and access and utilization of health care. Many of the questions were based on existing surveys (Behavioral Risk Factor Surveillance Survey, California Health Interview Survey, Canadian Community Health Survey and Mendocino Community Health Survey) and new questions were developed as needed to inquire about areas of rural health not previously explored, such as access to transportation, electricity, phones, computers and Internet as well as skills for responding to emergency medical situations and sources of health information.
The survey contained a combination of quantitative and qualitative responses. The quantitative responses were either categorical or continuous numerical answers. A 15% response rate was anticipated and it was desired to compare answers by county. Sample sizes were examined that would be necessary to satisfy a 10% level of precision (i.e., a bound on relative sampling error E of 10%), at a 95% confidence level. To satisfy this criteria, the desired sample size for each county ranged from n = 23 to n = 385 depending on the question. The number of surveys mailed took these estimates into account.
Surveys were mailed to a random sample of residents within the four county study site. The sampling strategy employed the use of a Geographic Information System (GIS) to map the population density for Zip Code Tabulation Areas (ZCTA) with an overlay of the locations of post offices. All of the post offices in low population density areas (<11 people per square mile) were selected (total post offices = 24; total post office boxes = 8165). Post offices located in higher population density areas (≥11 people per square mile) were randomly selected (total post offices = 19; total post office boxes = 15,441). Forty three different communities were sampled
Surveys were addressed to the box holder and the cover letter instructed someone in the household over the age of 18 to complete the survey and return it in the self-addressed stamped envelope. The survey was mailed to a total of 23,606 box holders.Surveys and follow-up reminder cards were sent during August and September, 2006.
Quantitative data was entered and analyzed using SPSS (14.0). To compare proportions and means Chi Square and ANOVA were used respectively for tests of statistical significance with a P value of <0.05 considered statistically significant. Differences found by Chi Square were explored using post hoc testing with Bonferroni adjustment to account for alpha inflation.
Qualitative data was entered and analyzed using ATLAS ti. Codes were developed for the answers to capture common themes.