Problem

The decennial Census Count has been completed end of 2020 and Census Bureau has begun the process of the 2020 Count Question Resolution (CQR) Program. North Carolina Office of State Budget and Management (NCOSBM) is responsible for the count resolution for the State.

NC State University Center for Geospatial Analytics sets out to assist NCOSBM with:

(1) identify areas where NCOSBM’s HU estimation model over- and under-count the Census 2020 housing units in place.

(2) where areas of under count occurs which may require adjustments in future population estimates and projections, identify potential causes that may help explain the phenomenon.

(3) develop a geovisualization application to be hosted on their website to communicate the findings to NCOSBM internal stakeholders, as well as the general public.

Results

The result from data observation points to the dominant factor of high percentage of African American and poverty rate for cities/towns with high under count. The result from geoprocessing shows the high under count areas (blue polygons) cluster on the outskirts of large cities: Charlotte, Raleigh, Greensboro. The Overlay tool executes fast and efficient when tested.

The resulting web application (Fig.2.) has eight widgets for users. The Info and FAQ to inform users on the function and step-by-step instructions. The Legend widget to display and List widget to turn feature classes on/off in the web map. The Overlay widget allows users to select the city/town and generates six demographic information, creating a profile for that city/town. The Filter widget lets users toggle between over and under count. The Chart widget lets users display numerical data in a bar chart, and the Print widget exports the results to the format selected by the user.

Analysis

There was a huge amount of data from the Census and NCOSBM's model for estimation. The original three excel files were split into different tables to join to six demographic feature classes at the data level: either Census Place or County level. This is done to increase efficiency of the Overlay geoprocessing (GP) service. This service was published from Model Builder Overlay (Fig.1) that intersects six demographic feature classes with the Over Count feature class. The six demographic feature classes: % African American, % Latin American, % Native American, % poverty, % housing vacancy, and % households without Internet. The Over Count feature class was created by the difference of two fields from Estimated housing units and Census 202 housing units feature classes. The Over Count feature class and six demographic feature classes were published to ArcGIS Online as a map service. The GP and map services' REST end points were used in the creation of the web application.

Reflection

Any issues arising from this project have been resolved before handing over to NCOSBM. Still, there are areas to consider for future updates. Additional features and tools can be developed to handle other Census related data such as population. The current Overlay tool can be further developed to allow users to select the demographic feature classes for study. Another feature can be developed to house spatial statistical methods as an advanced analysis tool.

This project highlights the importance of geospatial analysis, which is different from data analysis. The two analysis result in two very different findings, and thus two different perspective, giving rise to new approaches for tackling under count areas for 2030 Census. The project also test the processing speed of different workflow and arrive at the best time performance workflow. The last part of the project is to visualize the process in a Symposium Poster (Fig.3).