| Popis: |
Purpose: People with vision problems (VPs) have different needs based on their age, economic resources, housing type, neighborhood, and other disabilities. We used calibration methods to create synthetic data to estimate census tract-level community need profiles (CNPs) for the city of Richmond, Virginia. Design: Cross-sectional secondary data analysis. Subjects: Anonymized respondents to the 2015 to 2019 American Community Survey (ACS). Methods: We used calibration methods to transform the ACS 5-year tabular (2015–2019) and Public Use Microdata estimates into a synthetic data set of person-level records in each census tract, and subset the data to persons who answered yes to the question “Are you blind or do you have serious difficulty seeing even when wearing glasses?” To identify individual need profiles (INPs), we applied divisive clustering to 17 variables measuring individual demographics, nonvision disability status, socioeconomic status (SES), housing, and access and independence. We labeled tracts with CNP names based on their predominant INPs and performed sensitivity analyses. We mapped the CNPs and overlayed information on the number of people with VP, the National Walkability Index, and an uncertainty measure based on our sensitivity analysis. Main Outcome Measures: Individual need profiles and CNPs. Results: Compared with people without VP, people with VP exhibited higher rates of disabilities, having low incomes, living alone, and lacking access to the internet or private home vehicles. Among people with VP, we identified 7 INP clusters which we mapped into 6 CNPs: (1) seniors (≥ age 65); (2) low SES younger; (3) low SES older; (4) mixed SES; (5) higher SES; and (6) adults and children in group quarters. Three CNPs had lower-than-average walkability. Community need profile assignments were somewhat sensitive to calibration variables, with 18 tracts changing assignments in 1 sensitivity analysis, and 4 tracts changing assignments in ≥ 2 sensitivity analyses. Conclusions: This pilot project illustrates the feasibility of using ACS data to better understand the support and service needs of people with VP at the census tract level. However, a subset of categorical CNP assignments were sensitive to variable selection leading to uncertainty in CNP assignment in certain tracts. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. |
