Reply to Zhang et al.: Linear regression does not encapsulate the effect of non-pharmaceutical interventions on the number of COVID-19 cases

TL;DR

This paper critiques prior linear regression analyses of COVID-19 interventions, demonstrating that such models are inadequate for capturing the true effects of measures like mask mandates, and emphasizes the need for more appropriate modeling approaches.

Contribution

It highlights the limitations of linear regression in evaluating COVID-19 interventions and advocates for using differential equations like SEIR models for more accurate analysis.

Findings

Linear regression is insufficient for modeling COVID-19 intervention effects.

SEIR models provide a more appropriate framework for such analyses.

Prior conclusions about mask mandates may be unreliable due to model limitations.

Abstract

Zhang et al. (2020) used linear regression to quantify the effect of lockdowns on the number of cases of COVID-19. We show using differential equations from the susceptible-exposed-infected-recovered (SEIR) model and with an example from another location not previously considered that the Zhang et al. analysis should not be considered sound evidence that mask mandates are sufficient to control or the primary factor controlling the spread of COVID-19.