An empirical cross-sectional analysis of the corrections in the New York Times’ COVID-19 coverage

Background

To examine the errors and corrections in the New York Times (NYT) and to assess if there is an imbalance towards overstating the pandemic severity, which may support more extreme restrictions or understating the severity.

Methods

We conducted a retrospective cross-sectional analysis of COVID-19 articles that had corrections made and reported in the NYT “corrections page”. We categorized authors in a NYT COVID-19 article as a NYT reporter, NYT other, or independent author. We calculated the number and type of corrections by author type (NYT reporter, NYT other, and independent author) and number and percentage of corrections indicating an over- or under-statement of the COVID-19 situation.

Results

There were 576 total corrections for the included 486 articles. Forty-three percent (n=245) corrections specifically pertained to COVID-19. Compared to corrections not pertaining to COVID-19, corrections pertaining to COVID-19 were less likely to be about spelling (0% vs 23.6%), locations (1.2% vs 16.3%), or title/degree (0% vs 10.6%), and more likely to be about a vaccine/vaccination (21.2% vs 0.3%), incidence/cases of conditions (12.2% vs 0.3%), or disease testing (7.8% vs 0.3%; p<0.001).

Compared to corrections not pertaining to COVID-19, corrections pertaining to COVID-19 were less likely to result in an equivocal tone (16.7% vs 88.8%), but they were more likely to both overstate (54.7% vs 8.5%) and understate (23.7% vs 2.4%) the situation in the original text (p<0.001). Ten reporters (of 346) accounted for 24% of the corrections. The reporter with the single most corrections accounted for 7% of the corrections.

Conclusions

Differential tone of the corrections suggests bias in the reporting of COVID-19 topics in a top news outlet. The reporting of unbiased information is a first step in addressing issues of misinformation in public health messaging.