In a February 21st opinion piece published in the New York Times, columnist Bret Stephens critiqued mask mandates, writing in part that it was “widely observed” that “[s]tates with mask mandates fared no better against Covid than those without.” As we will show, this is demonstrably false.

Despite having worked on COVID-related issues for the past three years, we are actually unfamiliar with any empirical observations (based on data) of the sort offered by Stephens. But, having collected a very large state-level dataset on COVID-related attitudes and behaviors, we are well situated to evaluate the two key descriptive questions underlying his assertion:

  1. Did COVID-related death rates in states that had mask mandates differ from those in states that did not have mandates?
  1. Is there a difference in death rates between states with higher and lower levels of mask wearing?

In fact, these are actually pretty easy– descriptive– questions to evaluate. To answer them we zoom in on the key period when mask wearing was most important– between the end of the first wave (when there were no mask mandates) and the beginning of the vaccination campaign in the US: June 1 to December 31, 2020. We exclude the first wave, because it mostly occurred before any states adopted mask mandates; and we exclude the period after vaccinations became available, because vaccines would have been the most important driver of death rates after that point. That is, mask wearing, social distancing– all of the mandates that Stephens critiques– were the primary tools available to prevent the spread of COVID during this period of acute vulnerability.

To answer the first question, we plot the mortality rates from COVID (per 100,000 people) for this period in states that had mandates for all or most of this period and compare them to states that did not have mandates at all (11 states1) or had mandates for a minority of this period – less than 2 months out of the 7 month period under investigation (another 6 states2).

Figure 1

We see–in flat contradiction to Stephens’ assertion–that states without mask mandates (where 65.7% reported wearing masks) have an approximately 30% higher death rates than states with mandates, where 75.1% reported wearing masks (87.7 deaths per 100,000 vs. 67.4 per 100,000 in states with mask mandates).

Second, we evaluate the relationship of the rate of mortality rates against mask wearing during this period (averaged across the waves of our survey).

Figure 2

In short, we see a significant, negative, relationship between mask wearing and mortality rates. 

Correlation is not causation. Those states that had high levels of mask wearing generally had more stringent policies, as well as higher levels of social distancing and avoidance of crowds and public places. Further, there are certainly other, pre-existing, factors that influence COVID-related mortality numbers (like age of the population; density of housing, etc). Pulling apart that causal tangle is extremely difficult. But there is little doubt that mask mandates and mask wearing at the population level were associated with substantially lower levels of mortality. To claim otherwise in one of the most widely read and influential newspapers in the world, without evidence – indeed contrary to the actual evidence – is irresponsible, as it injects unwarranted confusion into an already fraught environment of rising distrust of public health institutions. 


1 The 11 states that had no mask mandates during the period under investigation are: Arizona, South Dakota, Idaho, Nebraska, Oklahoma, Missouri, Tennessee, Georgia, South Carolina, Florida, and Alaska.

2 Five of these six states (Iowa, New Hampshire, North Dakota, Utah, and Wyoming) first initiated mandates between mid-November and mid-December 2020, near the tail end of the time period we investigate, while a sixth, Mississippi, initiated a mask mandate on August 5, 2020, but then canceled it the next month, on September 30th.

Read journal articleRead reportLearn more