Interactive Impact Labs

Modeling US Communities

Client Challenge: In 2018, a prominent foundation launched an initiative to promote upward mobility and economic opportunity for low-income youth. In order to identify where to invest significant research time in the form of qualitative interviews, the team needed a data-driven way to compare the lived experiences of different geographic communities across the entire United States.

Client

CONFIDENTIAL

Category
Data aggregation,
Geospatial modeling,
Racial Equity focus
Delivered

December 2018

ii Labs Solution:

Carefully combining over 70 national datasets, we developed an interactive map that layered descriptive data across indicators of health, economic well-being, socio-economic factors, demographics, education, and crime. In doing this, the team was able to select several geographies where they would sample and eventually conduct in-depth interviews. Results from these interviews were used to better understand how low-income youth, particularly Black and Latinx adolescents, perceive their available educational opportunities.

Data Methodology

We grouped these 70 indicators into a conceptual model of four thematic indices: the (1) Health in the community, (2) Economic opportunity in that place, (3) Neighborhood strength in that place, and (4) Educational strength of the community.

Each of these four broad themes also had 2 or 3 sub-indices, such as the difference between (3.1) Housing security, (3.2) Racial segregation, and (3.3) Violence/deaths of despair… all of which play a role in the strength of a neighborhood.

While index scores are incredibly useful for analyzing and understanding difference, we recognize the inherent over-simplification of complex social phenomena when flattened to a set of numbers. We therefore proceeded according to a set of best practices: All of the data needed to be normalized, calculated against its standard deviation (z-scoring), and made uniform in the direction of positive vs. negative numbers (for more desirable outcomes of that measurement vs. less-desirable ones).

The project for this client chose not to double-weight certain data measures over others, in the absence of any strong justification for doing so.

The result of this analytical model was the unique ability to compare outcomes, relative to each other, in any and every locality in the 48 continental States. For example, any particular locale—whether urban, suburban, or rural—might be experiencing better outcomes in Economic activity, while falling behind in Health, Neighborhood, and/or Educational strength.

To make these comparisons as simple as possible for the foundation to use, we rescaled the four Component Scores onto a scale of 0–25, where 25 would indicate those places experiencing the most beneficial measurements amongst the whole country.

Adding these four scores together, it also produced one overall Composite number on the scale of 0–100. While this is never the end of a truly holistic analysis, it becomes an extremely useful, reusable tool for researchers to rapidly identify places to investigate further.