Public health inequalities, structural missingness, and digital revolution: time to question assumptions
November 22, 2021
Elena N. Naumova
J Public Health Policy. 2021 Dec;42(4):531-535. Epub 2021 Nov 22.
PMID: 34811463 | PMCID: PMC8607058 | DOI: 10.1057/s41271-021-00312-y
Over my career in public health as a data scientist, I increasingly encountered a persistent problem: populations that need the most attention of public health professionals are the populations about which we have the least information. People need attention because of their ongoing health concerns or risks and hazards imposed by society and the environment. The information, data, and thus the knowledge about these groups are often incomplete, inconsistent, outdated, and simply do not exist. And the data ‘missingness’ is something we systematically avoid discussing in publications, data analyses, algorithm development, policymaking. Our common assumption about missing data is that their effect on our scientific findings and practical decisions is inconsequential. We typically assume that research designs, statistical tools, logical analysis, and technological advances are immune to data flaws. We often assume that the missing data are rare and random. Or, if not, that we are well equipped to handle this problem.