Dartmouth study evaluates tone of Ballotpedia’s COVID-19 coverage


Earlier this year, a team of researchers at Dartmouth developed a method to evaluate the tone of news articles written about Covid-19. Using standard sentiment indicators, the Dartmouth team assessed the text of 42,000 articles from dozens of U.S. and international news sources.

The study found that U.S.-based news sources used significantly more negative language than the international outlets studied. 87% of major U.S. media stories were negative in tone compared to 50% of non-U.S. sources. The international sample included articles from the U.K.’s BBC, Canada’s CBC, and Australia’s ABC; each country’s dominant news source, all publicly owned. The U.S. outlets analyzed in the study are privately owned with more competitors within the country’s borders than the international outlets studied. The political leaning of the outlets made no difference in their usage of negative language.

The Dartmouth research team cross-checked the negative sentiment indicators of articles in the international and U.S. press against the extent of COVID spread in the relevant countries. They found that there was no correlation between the number of COVID cases in a particular country and the extent of negative tone in the country’s news sources.

Ballotpedia submitted several hundred articles from its coverage of Covid-19 to find how it would compare to the articles in the study. The Dartmouth team found Ballotpedia used significantly fewer negative sentiment indicators in its language choices compared to the major U.S. media outletsBallotpedia articles were .92 standard deviations less negative than the average of all other sources analyzed.