Undergraduate Research Assistantship

Undergraduate Research Assistant in the Department of Geography and Planning Appalachian State University, Boone, NC

Within the Department of Geography and Planning, I have had the opportunity to work on a semester-long research project alongside a professor through a grant from the Office of Student Research. For the Spring 2020 semester, I was selected to work with Dr. Maggie Sugg, Lauren Andersen, and Dr. Elizabeth Shay.

This Undergraduate Research Assistantship (URA) project is focusing on an extension of the Social Determinants of Health (SDOH) project, in addition to completing complex spatial analysis and statistical analysis. The objective of this study is to map and analyze the spatial distribution of SDOH across North Carolina at a sub-county level. To assess SDOH, between 50 to 100 variables have been collected from multiple sources, including the American Community Survey (ACS), North Carolina Center for Geographic Information and Analysis, and Homeland Infrastructure Foundation-Level Data (HIFLD). The SDOH variables encompass Employment, Education, Household Conditions, Household Income, Work Transportation, Health Factors, and the distribution of public and health services. A principal component analysis (PCA) is being used to address multicollinearity among variables and aggregate data into components. Geographic Information System (GIS) based methods will be used to examine heterogeneity and identify locations that require targeted public health interventions to address underlying health disparities.

New Processes - What I Have Learned

During my studies at Appalachian State, I have completed a Writing within the Discipline course where I learned how to write a scientific abstract. I applied that knowledge, and further learned how to write a successful and informative abstract based on this URA project. This abstract was submitted for the opportunity to present in a Student Research Presentation at Appalachian State. Unfortunately, due to the developments of the COVID-19 pandemic, the event was canceled; however, the experience of collaborating and writing an abstract is a skill I will carry on into my future work.

Abstract written for URA detailing processes researching SDOH

As I continue this URA, I have also learned how to use IBM SPSS to compute a PCA. It is my first experience with SPSS, and I am learning the value of computing this type of statistical analysis to understand the factors most important to map.

These images illustrate the results from completing a PCA, as the components created will be used to map. The collected data from American Community Survey data was first standardized into Z-scores. A number of processes with SPSS such as Bivariate Correlation and Dimension Reduction analysis were all used to generate the PCA results. Kaiser's Criterion is used as the determining factor of which components will be used. This determination is all components with eigenvalues greater than one. The Scree Plot illustrates the components meeting Kaiser's Criterion or those with eigenvalues greater than one. These products were both generated from completing the analysis in SPSS. This PCA includes nearly 65 variables that consist of the SDOH data, in addition to Health Professional Shortage Area (HPSA) data from a previous study of Dr. Maggie Sugg, acquired from the Health Resources and Services Administration.

SPSS developed PCA analysis and generated components from a PCA analysis.

A scree plot that illustrates the distribution of eigenvalues from the SPSS PCA analysis.

From these results and statistical analysis, components were named and cardinality was determined.
Cardinality represents factors that either decrease overall vulnerability, ( - ), or increase overall vulnerability ( + ).
After determining components with eigenvalues greater than one and determining cardinality,
work within a GIS to create map outputs is possible.

PCA Chart of Components, Cardinality, Names, and Variables

Component 1 through Component 8 are listed in the chart with the variance, top 5 variables, and cardinality.

PCA Chart of Components, Cardinality, Names, and Variables

Component 9 through Component 16 are listed in the chart with the variance, top 5 variables, and cardinality.

PCA Chart of Components, Cardinality, Names, and Variables

Component 1 through Component 8 are listed in the chart with the variance, top 5 variables, and cardinality.

1/2

Anchor 2

Preliminary Maps

The following maps were created by primarily using ArcGIS Pro. These maps are my second experience with extensive mapping in ArcGIS Pro. Most of my experience is by using Esri's ArcMap with my coursework exercises. I enjoyed learning how to use ArcGIS Pro during this process. I have gained experience in this Esri platform, where I feel confident mapping within ArcGIS Pro and ArcMap.

I created approximately 10 different maps in different layouts in ArcGIS Pro.

Additionally, before mapping, the components were used to create a Vulnerable field. Based on the cardinality of the components, the Vulnerable field includes them either by addition or subtraction to generate this field of vulnerability.

The images below include maps of the following topics:

Vulnerability of SDOH from the first PCA results using the Vulnerable field
Vulnerability of SDOH from the second PCA results using the Vulnerable field
Vulnerability Map with PCA Component of Poverty by North Carolina Regions from the first PCA
Vulnerability Map with PCA Component of Poverty by North Carolina Regions from the second PCA

Vulnerability of SDOH from the first PCA results using the Vulnerable field

Using ArcGIS Pro and Cartographic Boundary Shapefiles, Vulnerability across Census tracts is displayed for the state of North Carolina.

Vulnerability Map with PCA Component of Poverty by North Carolina Regions from the first PCA

Using ArcGIS Pro and Cartographic Boundary Shapefiles, Vulnerability across Census tracts is displayed for the state of North Carolina. Additionally, from the first PCA, Component 1 of Poverty is shown by Census tracts as well.

Vulnerability Map with PCA Component of Poverty by North Carolina Regions from the second PCA

Using ArcGIS Pro and Cartographic Boundary Shapefiles, Vulnerability across Census tracts is displayed for the state of North Carolina. Additionally, from the second PCA, Component 1 of Poverty is shown by Census tracts as well.

Vulnerability of SDOH from the first PCA results using the Vulnerable field

Using ArcGIS Pro and Cartographic Boundary Shapefiles, Vulnerability across Census tracts is displayed for the state of North Carolina.

1/4