The Iowa Caucus was a huge mess for a number of reasons, one of which may be that the credibility of news coverage of the Democratic candidates had a significant drop in the days before and during the caucus. While it’s unclear what impact this may have had on results in Iowa, it suggests that as news consumers we should be more circumspect this week in New Hampshire, and indeed throughout the election season.

How do we know news credibility dropped?

The data gathered by the Factual’s credibility algorithm allows us to broadly track the trends in article credibility across the entire media ecosystem. In short, the algorithm can instantly assess the credibility of a news article by looking at where an article was published, who wrote it, how many unique and reliable sources were used, and how opinionated the writing’s tone is. This grade is fed into, and informed by, our database of more than 7 million graded articles. (For more on how this works, see our recent post.) The resulting data gives us the tools to assess whether news coverage actually declines around election time, and we decided to begin investigating this phenomenon with the Iowa Caucus.

Want to spend less time searching for the best news stories?
Get the best news in your inbox every morning. Determined by data, not politics.
Thank you!

Please check your email for instructions to ensure that the newsletter arrives in your inbox tomorrow.

Oops! Something went wrong.

Our data suggests a worrying story.

The credibility of news articles related to the Democratic candidates decreased substantially in the days before, after, and during the Iowa Caucus. While the average credibility of articles is around 62.6% in our data on coverage of the Democratic presidential candidates from October 1, 2019-January 20, 2020 (2 weeks before the Iowa Caucus), that average declined substantially in the days before the Iowa Caucus, and was particularly low the week of the caucus.

Screen Shot 2020-02-13 at 12.37.06 PM

What does the decline in average credibility tell us?

The credibility of articles during the week of the Iowa Caucus plummeted to 55.7%, almost 7 points lower than our average over October 1-January 20; 10 of the 12 worst scoring days occurred within 2 weeks of the caucus. This result is also statistically significant at the 95% level, meaning we are 95% confident this is not due to random variation in the data. This suggests that the media environment broadly trended toward less credible news coverage around the time of the contest.

While this can be partially explained by the confusion around the results and the desire of the media to cover every twist and turn, our data also shows that the daily average had reached 55.8% the day before the caucus. Moreover, data from the week before the caucus (January 26 – February 1) shows that entire week’s average had already fallen to 57.7%. That means that the credibility of news articles about the candidates, for whatever reason, was substantially lower than normal well before the proverbial train went off the tracks in Iowa.

This decline in credibility was also apparent across the political spectrum.

In our dataset, outlets with an orientation right of center have tended to deliver less credible coverage of the Democratic candidates (scoring an average of 57%). But outlets on the Right and the Left saw their scores drop by a similar degree before and during the Iowa Caucus.

The week before the Iowa Caucus, average scores dropped anywhere between 2.2 points (Moderate Right) and 6.7 points (Right) off their overall averages. This trend intensified the week of the caucus, with all scores dropping off their averages by anywhere between 5.7 points (Moderate Right) and 7.7 points (Right).

This also means that the average credibility of news articles about the candidates was at least 5 points lower the week of the Iowa Caucus, regardless of where you looked on the political spectrum.

Partially, this dispels any notions of election-related misinformation being the purview of a particular part of the political spectrum. It might seem like outlets on the Right have the most to gain from spreading misinformation during the caucus — using confusion and conjecture to sow disunity on the Left — but Moderate Left outlets actually saw the second biggest decline (7.3 points) in credibility during that week. And while Moderate Left outlets may have the highest overall average from October 1-January 20 (67.3%), this data suggests that any perceived level of journalistic rigor is not impervious to the demanding dynamics of election scenarios.

Restore your confidence in the news.
Get the best news in your inbox every morning. Determined by data, not politics.
Thank you!

Please check your email for instructions to ensure that the newsletter arrives in your inbox tomorrow.

Oops! Something went wrong.

What does this mean for you?

The first takeaway is that the public needs to be on their guard this election season, especially in the time leading up to and around major election events, such as primaries, major debates, and the general election on November 3. Regardless of where you get your news, it’s likely to be not as well substantiated as usual, meaning we all have to put greater due diligence in before believing or sharing news articles about the election, especially because we now know lies spread faster than the truth.

It is less clear how strong a connection we can make between our article credibility scores and the active, intentional spread of disinformation. For example, that articles are 5% less credible or more in a certain time period does not necessarily mean they are false; rather the complexity or time constraints associated with specific events may lend themselves to articles with fewer sources or opinionated writing. For example, outlets covering Iowa were left to speculate about what went wrong (USA Today – 58%), who was responsible (The Independent – 59%), and when we would know the results (The Washington Times – 47%), all in an environment of limited knowledge. In that case, it’s no surprise that article credibility took a hit.

We intend to continue to investigate this trend through the 2020 primaries and especially in the run-up to the general elections to see if misinformation looks to be as large of a factor this year as it was in 2016.

You can keep track of article credibility on your own by downloading our Chrome extension to assess article credibility in real time. If you prefer to have the best news delivered to you every morning, subscribe to our daily newsletter. It is backed by The Factual’s credibility algorithm so that you only receive the best-resourced articles from across the political spectrum.

* Note: Some figures have been updated since February 4.

Epilogue: Article Credibility Trend Continues in New Hampshire

The news articles published just before and after this week’s primary in New Hampshire showed a continuation of the declining average credibility trend that we observed last week in Iowa. This suggests that the statistically significant drop we saw was not an anomaly resulting from the contested results in Iowa but rather a general observation that news around election events is less credible than normal.

Screen Shot 2020-02-13 at 4.26.17 PM

This trend makes intuitive sense. News outlets know that articles on elections get a lot of traffic, which raises the publication frequency in the days and weeks around election events. But many of these are thinly researched and overly opinionated in order to stand out. The result is a lot of unsubstantiated ideas that may waste your time.

We’ll monitor this trend throughout the 2020 U.S. election season and, as always, highlight the most credible stories across the political spectrum on every candidate and issue.

Published by Phillip Meylan

Phillip is a writer, researcher, and editor. At The Factual, he leads research efforts that utilize the company's ever growing data on the media ecosystem. He is also a contributor to FP Analytics, Foreign Policy's research and advisory division, and an adjunct fellow with the Center for Strategic and International Studies.