Measuring the Politicization of Official COVID-19 Death Statistics

• In Event: COVID in the United States

Abstract

The COVID-19 pandemic resulted in an unprecedented level of data collection and dissemination by state and non-state actors. The relative availability of this data and the salience of the pandemic catalyzed abundant social scientific research attempting to relate political regime, political partisanship, or public health policy to COVID-19 outcomes. I argue that the use of COVID-19 outcome measures as a dependent variable in this kind of analysis is misguided, because this data is produced by political actors and thus is endogenous to the political systems that produce it. To be more concrete, it could be that, as some political scientists argued early in the pandemic, there are fewer COVID-19 cases and deaths in authoritarian regimes. Conversely, it could also be that there are actually more COVID-19 deaths in authoritarian regimes, but fewer reported deaths if, for example, it is easier or more incentivized to intentionally undercount deaths in authoritarian regimes.

I create an identification strategy that can be used to measure this politicization of COVID-19 death data, and apply the measure at the state level in the United States. I define the measure as the deviation between COVID-19 deaths reported officially by a government and the excess death count, operationalized by (reported COVID-19 deaths/(excess deaths)*100). Excess mortality measures the deviation of deaths from all causes in a given year from an expected number of deaths given historical averages; during a pandemic, this can offer a counterfactual. Larger underestimates of official statistics compared to excess mortality can be attributed to the successful political manipulation of those official statistics. Preliminary analyses in a US context show that Republican states undercount COVID-19 deaths to a greater degree than Democratic states— that COVID-19 death data is endogenous to the partisanship of the state.

The production and dissemination of government COVID-19 statistics is a political process, and thus should be used with caution by researchers. This is salient in both political science and public health beyond the geographic and pandemic context, contributing evidence to the theory that data is endogenous to politics and providing an example of how one might conceptualize and measure the politicization of data.

Author

• Rachel Kulikoff, University of Michigan