Upgrade to Pro — share decks privately, control downloads, hide ads and more …

stitching-across-data-seams-stlc

 stitching-across-data-seams-stlc

Christopher Prener

August 12, 2021
Tweet

More Decks by Christopher Prener

Other Decks in Research

Transcript

  1. Stitching Across Data Seams Tracking COVID-19 Disparities in Missouri Christopher

    Prener, Ph.D. Assistant Professor of Sociology, SLU 08.12.2021
  2. Tracking COVID-19 Disparities in Missouri Christopher Prener, Ph.D. Assistant Professor

    of Sociology, SLU 08.12.2021 Stitching Across Data Seams
  3. Acknowledgments Saint Louis University Department of Computer Science Especially Chair

    Michael Goldwasser, PhD and David Ferry, PhD Student Contributors: Alvin Do, Metta Pham, and Eric Quach
  4. AGENDA 1. Preface 2. How can the public find information

    on COVID? 3. What is wrong with this arrangement for COVID data? 4. What does the sewing kit look like for my COVID work? 5. What COIVD disparities exist in Missouri and St. Louis? ST. LOUIS COVID-19 COMPARATIVE MODELING NETWORK | 08.12.2021
  5. ▸ Medical and urban sociologist with an interest in spatial

    and computational methods ▸ Affiliations: • SLU’s Institute for Healing Justice and Equity • Northeastern University’s Institute for Health Equity and Social Justice Research • Core Faculty, SLU’s Advanced HEAlth Data Research Institute 1. PREFACE “HI, I’M CHRIS”
  6. The AHEAD Institute underscores the University’s commitment to addressing public-health

    issues and improving patient health outcomes — especially those that disproportionally affect poor and underserved communities — in alignment with the University’s Jesuit mission. To achieve this commitment, the AHEAD Institute relies on health care data for its truly remarkable potential to create new efficiencies for physicians and more positive outcomes for patients. Leslie Hinyard, Ph.D., MSW Executive Director [email protected] Jeffrey Scherrer, Ph.D. Senior Director for Research [email protected]
  7. 1. PREFACE PUBLIC SCIENCE “presenting findings in an accessible manner”

    engaging in descriptive research that moves public discourse forward
  8. 3. TRACKING A PANDEMIC DATA ARE OFTEN POORLY VISUALIZED Basic

    rules, like using per capita rates, are often ignored.
  9. DIFFERENT TOOLS Dashboards are being powered by a number of

    different commercial tools, including ESRI, Microsoft, Tableau, and in-house solutions.
  10. DIFFERENT TOOLS Dashboards are being powered by a number of

    different commercial tools, including ESRI, Microsoft, Tableau, and in-house solutions.
  11. DIFFERENT TOOLS Dashboards are being powered by a number of

    different commercial tools, including ESRI, Microsoft, Tableau, and in-house solutions.
  12. DIFFERENT TOOLS Dashboards are being powered by a number of

    different commercial tools, including ESRI, Microsoft, Tableau, and in-house solutions.
  13. DASHBOARDS ≠ OPEN DATA Few dashboards provide easy access to

    underlying data, though it is there if you know where to look. There is also little to no standardization.
  14. DASHBOARDS ≠ OPEN DATA Few dashboards provide easy access to

    underlying data, though it is there if you know where to look. There is also little to no standardization.
  15. Sewing these seams requires an array of computational tools for

    scraping and standardizing various jurisdictions’ data, plus a communications strategy. Unsplash
  16. 4. THE SEWING KIT PULLING THE FABRIC TOGETHER New York

    Times COVID-19 Database (via GitHub) County Public Health Zip Code Data (via scrapers+API calls) Missouri COVID Tracking Data Sets State of Missouri and Illinois (via scrapers) CMS Nursing Home Data & HHS Hospitalization Data (via API) Census Bureau (via API)
  17. Warren County stopped reporting ZIP code data at the end

    of June Estimated ZIP code populations and issues with data reporting
  18. FOR THE PROBLEM OF THE TWENTIETH CENTURY IS THE PROBLEM

    OF COLOR LINE W.E.B. Du Bois The Souls of Black Folk (1903) Wikimedia Commons
  19. March-June: r = -0.736 (p < 0.001) July-September: r =

    0.047 (p = 0.718) October-December: r = 0.642 (p < 0.001)
  20. March-June: r = 0.644 (p < 0.001) July-September: r =

    -0.063 (p = 0.628) October-December: r = -0.633 (p < 0.001)
  21. 6. FINAL THOUGHTS REFLECTIONS ▸ Research, data, and communication are

    not the first priority: ▸ COVID-19 data sources suffer from a lack of standardization and often cannot be accessed. ▸ COVID-19 “dashboards” are all the rage, but there are frequent issues across jurisdictions with how similar metrics are being communicated. ▸ COVID-19 patterns that appear durable as cross-sections have important period effects that we need to interrogate. ▸ Iterating on analyses is not something we always get to do, but it is tremendously gratifying. ▸ Open data and science are also about community and communication.
  22. Slides available via SpeakerDeck Follow on the web: speakerdeck.com/chrisprener/ stitching-across-data-seams-stlc

    Raw data, code available via GitHub github.com/slu-openGIS/ MO_HEALTH_Covid_Tracking [email protected] chris-prener.github.io LEARN MORE THANKS FOR COMING! @chrisprener Visualization code available via GitHub github.com/slu-openGIS/ covid_daily_viz slu-opengis.github.io/ covid_daily_viz/ chrisprener.substack.com