Reducing uncertainty in the american community survey through data-driven regionalization.

Area of Science:

• Social Sciences
• Geographic Information Systems
• Statistical Analysis

Background:

• The American Community Survey (ACS) is the primary source for US neighborhood-level demographic and economic data.
• ACS estimates are frequently unreliable, with high margins of error impacting policy and research.
• Over 72% of census tracts have margins of error exceeding estimates for children under 5 in poverty.

Purpose of the Study:

• To present a heuristic spatial optimization algorithm for regionalization.
• To demonstrate reducing margins of error in ACS data through composite geographies.
• To achieve user-specified thresholds for data reliability.

Main Methods:

• Development and application of a heuristic spatial optimization algorithm.
• Implementation of a regionalization process to create new composite geographies.
• Utilizing an open-source algorithm for processing survey data.

Main Results:

• The regionalization algorithm successfully reduces margins of error in survey data.
• Composite geographies are created to enhance data reliability.
• The method allows for achieving user-defined error reduction targets.

Conclusions:

• Regionalization offers a viable solution to improve the reliability of ACS data.
• This approach enhances the usability of survey data for policy, research, and governance.
• The open-source algorithm provides a practical tool for data uncertainty reduction.