ABSTRACT
This article attempts to solve a puzzle regarding the sensitivity of analysis results of COVID-19 infection and fatality. Our findings suggest that measurement errors, statistical significance of explanatory variables, and regional differences play a crucial role in the sensitivity of results. The significance of political stability, governance indicators, medical resources, demographic features is inconsistent and depends on the source of data, the choice of the time period, and region. This article also provides evidence that careful data screening and use of the moving average technique can alleviate the sensitivity issue and produce fairly robust results. We conclude that social science research on COVID-19 should not underestimate the issue of data quality and should refine raw data to minimize random error. If the sources of measurement error are not carefully managed and intensive statistical tests of sensitivity are not verified, data quality will end being subjected to the skepticism of evidence-based policy making.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
2 Our World in Data, https://ourworldindata.org/covid-sources-comparison.
3 JHU, https://coronavirus.jhu.edu/.
4 France, COVID-19 dashboard, https://dashboard.covid19.data.gouv.fr/.
5 Palestine government website, http://site.moh.ps/index/covid19/LanguageVersion/0/Language/ar.
6 Ministry of Health of Kazakhstan, https://www.coronavirus2020.kz/ru.
Additional information
Notes on contributors
Yeobin Yoon
Yeobin Yoon is a doctoral student at the School of Public Affairs, Penn State Harrisburg.
Bum Kim
Bum Kim is a researcher at the Asia Regional Information Center, Asia Center, Seoul National University, and a doctoral student at the Graduate School of Public Administration, Seoul National University. His research interests are in public choice, policy analysis, and evaluation.