The Connecticut Data Collaborative (CTData) explored 500 Cites data and local data to examine the relationship between health and housing and understand the challenges faced by Hartford neighborhoods struggling most from disinvestment. As a result, the CTData team created two indices—the Housing Conditions Index and Housing Stability Index—that allow the field to compare health outcomes with housing conditions and housing stability at the census-tract level. With these data, the team conducted an optimized hotspot analysis to identify significant clusters within neighborhoods needing public investment and employed a data dissemination strategy to help those using public data improve their data literacy.
HOW THEY USED 500 CITIES DATA
CTData created the Neighborhood Data Explorer, an online data platform that gives the public access to all the data used in the analysis and features many of the 500 Cities data health outcomes. Through interactive visualizations, users can see the neighborhood conditions for health, economic, and demographic indicators and how they overlap with housing stability and conditions. Users are also provided information about the indicators and their importance.
The Connecticut Data Collaborative also created Health in Hartford’s Neighborhoods, an interactive online story that walks users through the CTData team’s analytic approach to identifying the relationships between housing conditions, health outcomes, and neighborhood disparities. This webpage offers an overview of how the team used the 500 Cities data, as well as an overview of the two indices created for the project.
The CTData team constructed these two indices using publicly available data. The Housing Stability Index examines housing finances and tenure by measuring occupancy, the rent-to-income ratio, the mortgage-to-income ratio, eviction rates, foreclosure rates, average length of tenure, and assessed price per square foot. Alternatively, the Housing Conditions Index examines aspects of housing quality related to the physical quality of housing stock as part of the built environment by measuring housing code violations, vacancy rates, and fire incidents.
In their analysis of these two indices and 500 Cities data, the CTData team found that someone living in a highly unstable tract was 34 percent or 36 percent more likely to report being in poor mental or physical health than someone living in a tract with a high housing-stability score. The team published a final report that outlines their methodology and findings for the field.
The Connecticut Data Collaborative employed several methods for sharing their results both in the earlier stages of analysis and after finalizing key findings. Those methods included the following:
- Sharing preliminary results at a quarterly meeting. CTData hosts a quarterly data meetup in Hartford for local data professionals. At one of these meetings, the team shared preliminary results via storytelling maps that engendered a robust dialogue about other public data to include in the analysis and ultimately informed the final analysis. The meeting engaged 45 professionals representing various sectors including nonprofits, universities, state agencies, municipal employees, and community foundations.
- Sharing final results via online meetings. Once their results were finalized, CTData held a series of online meetings, guiding the audience through the online data story. They also held a series of in-person meetings with neighborhood organizations to share the data and the data-literacy curriculum developed to teach folks about using public data.
- Sharing final results via a data walk. CTData's final meeting was a participatory event where people were divided into groups and led through a data walk. CTData displayed the three most compelling maps and asked group members to engage with and comment on the data being displayed.
Finally, the CTData team created a guide to creating indices that provides a step-by-step approach for policymakers, planners, community-development workers, and others looking to create an in-depth system for using data to evaluate housing. This resource outlines how other communities can advocate with public data by creating indices that can answer specific questions and put local data to work.
WHAT THEY LEARNED
Throughout this process, CTData offered different opportunities to explore the results in an attempt to reach the broadest possible audience. Their goal was to engage folks from all sectors with diverse areas of expertise, as well as the residents and community the analysis focused on.
By opening multiple avenues of engagement, they were able to build regional advocates' knowledge, equip funders with crucial information, and provide residents with research about their neighborhoods.
In short, they learned that big players in the region were using their findings to spark action.
- A Hartford-based community foundation remarked that the data and results that CTData provided were being used to determine where targeted investment in the community could have the greatest impact.
- CTData also heard from several advocates who were using these data in their grant proposals. The results of their efforts unearthed new information and provided the insights necessary to initiate action.
The Connecticut Data Collaborative team also learned key valuable lessons while working with the 500 Cities data. When using estimates like 500 Cities data as a dependent variable, the CTData team found that it is important to account for the measures that went into determining the estimates. That is, it does not make sense to include variables such as race or income on the right side of a regression equation if those variables are incorporated in the health estimates.
Moreover, it is important to interpret results carefully if there are strong correlations between the variables being tested (such as foreclosures, in this case) and a variable used to create the 500 Cities health estimate (such as race or income).
WHAT THEY RECOMMEND FOR SIMILAR PROJECTS
The Connecticut Data Collaborative has identified the following recommendations for organizations interested in pursuing similar approaches to using the 500 Cities data in their communities:
- Working with public data is complex. It is not always available or readily suitable for analysis. In our project, we relied on federal census data and local Hartford data available on the city’s open data portal. One of the datasets on the Hartford data portal that we used includes reports by residents of housing code violations, and these data are updated nightly. However, because of the coding in the system, we were not able to discern whether claims were substantiated, many claims were classified under a generic code, and the city was not able to implement a viable solution within our study's time frame. Such challenges are common when using data in general, and are often not identified until you start delving into the details. Although we were not able to analyze the entire dataset, we gleaned important information from it, and it yielded highly significant findings for our work. We would like to revisit this area with the city and see if, over time, the data’s coding could be changed to provide more fruitful information and allow us to conduct a more extensive analysis.
- An unexpected benefit of this work was the development of new partnerships. Organizations that were not previously in our network are now aware of our work and actively engaged in the opportunities we presented to share the progress of our study. The Asthma Center at Connecticut Children’s Medical Center reached out to find ways to integrate their data in our analysis. However, because of the timeline of the IRB process and HIPAA requirements, we could not incorporate it into this project, though our discussions about including it in the future are continuing.
- People enjoyed engaging and being involved in a dialogue about the work. There are not many opportunities for people to gather and engage with a broad cross-sector audience and confer with others. During our final event (which included a data walk and discussion) people enjoyed hearing from others, sharing their feedback, learning about the work, and engaging in a dialogue. A report is great, but interactively sharing results was a more fruitful experience for our audience and provided tangible next steps and ideas for our organizations.
To learn more, follow @CTOpendata, @TrinityCollege, and @ActionLabHtfd on Twitter. If you are interested in learning more about this team’s project approach or have specific questions regarding replicating their work in your community, please feel free to contact the Principal Investigator of this project, Michelle Riordan-Nold.