Research Methods

Selection of Study Area

The focus of this project is on two counties out of 31 potential counties in Northeastern North Carolina.  Tier 1 counties, defined as the 40 most economically distressed counties by the NC Department of Commerce (2013), are of special interest.  North Carolina determines rankings based on an assessment of each county’s unemployment rate, median household income, population growth, and assessed property value.  In addition, any county with a population of less than 12,000 or a county with a population of fewer than 50,000 residents with 19 percent or more of those people living below the federal poverty level automatically is designated among the most distressed counties. We selected two counties of interest to narrow the scope of the project. Based on available data on Tier 1 Counties, we decided to focus on Beaufort County and Washington County.   We chose Beaufort in part to leverage connections Resourceful Communities and other community organizations have in the county, and because of its applicability to the interests of other economic development organizations (for example, Carolina Farm Stewardship Association, which is pursuing a small grants program there).  Though Beaufort county includes some extremes of poverty and wealth (i.e., the town of Washington), it represents the situation of rural, Caucasian farmers in eastern North Carolina very well.  Washington County was chosen as a second research area as it is one of the poorest counties in North Carolina, and because its agriculture and service sector population demographics are primarily African-American, but are otherwise similar to Beaufort County.

Notably, Washington County is much smaller than Beaufort County, with a population of 12,736 in 2012 (Census Bureau, 2014) as compared to Beaufort County’s 47,507 (Census Bureau, 2014).  Beaufort County is primarily white (71.7%) while Washington County is split between Caucasian and African-American residents (47.9% and 49.7% in 2012, respectively).  Beaufort County also has a relatively higher median household income ($40,147 in 2012) versus Washington County ($33,718 in 2012). The poverty rate is 26.5% in Washington County versus 20.6% in Beaufort County in 2012.  Beaufort County and Washington County are similarly educated (80.6% with a high school degree in Beaufort County versus 80.8% for Washington County) (Census Bureau, 2014).

Methodology and Programs

ArcGIS 10.1, a spatial analysis program, was used to analyze existing USDA Food Desert data and to identify food deserts in the study area that do not meet the USDA definition but still suffer from food inaccessibility.  Yiduo Zhang mapped the location of different producers and food hubs, including traditional grocery stores, farmer’s markets, produce stands, and convenience stores.

The statistical analysis was informed by the Food Environment Atlas, developed by the USDA Economic Research Service to provide county-level data on variables affecting food environments, including economic, demographic, and local food access variables.

The survey and interview methodology was informed by the community food assessment tools developed by USDA Economic Resource Service and refined by the Virginia Cooperative Extension, a previous assessment of Beaufort County developed by Jared Cates of Carolina Farm Stewardship Association, and a review of food supply chain analysis literature in the rural South.  USDA Agricultural Research Service (ARS), US Census Bureau data, and North Carolina economic data will be the primary sources of information.

Private, one-on-one interviews and limited retail and consumer surveys were the primary methods by which data was collected.  This required institutional review board (IRB) approval, received in November 2013.

The interview data gathered was transcribed using professional transcription software InqScribe.  NVivo, qualitative data analysis software available for free to Duke students was used to identify themes in the interview data. As some yes or no questions and quantitative questions were asked, STATA, which is only available at a subscription cost to Duke students, and Excel 2010 were used to graph and define statistics of the interview populations.