How geography, headlines and fact-checking affect campaign coverage engagement

State and local campaign coverage sees greater time on page than federal race coverage

Each story was classified based on whether it focused on a federal race (U.S. Senate or House) or a state-based race, including both statewide (e.g. attorney general) and local (e.g. mayoral) contests.

Articles referencing state-related campaigns had a higher average time on page than articles that did not reference these types of campaigns. There was no significant effect of the type of race on the number of page views or social referrals the article received.[ref In an OLS regression model predicting average time on page with a news story (with controls for the news outlet, article word count, type of article, headline type, strategy mentions, and issue mentions), state-based political campaigns significantly increased the time on page (B=.25 SE=.12; p<.05). In two regression models predicting story page views and social referrals (with controls for the news outlet, article word count, type of article, headline type, strategy mentions, and issue mentions), state-based political campaigns did not influence the number of page views (B=.17 SE=.15; p=.28) or social referrals (B=.08 SE=.23; p=.72). The dependent variables – page views and social referrals – were modeled using negative binomial regression.] As the graph below illustrates, an article referencing a state-based political race increases read time by 14.7 seconds.

[chart slug=”electioncoverage-6″]

Election articles with clickbait headlines have fewer page views than traditional headlines

The headline on each election news story was examined for whether it was a traditional headline or a question-oriented headline (sometimes called “clickbait” in more popular press, but we use it here in a neutral manner to mean a headline which attempts to create intrigue by posing a question or leading the audience to read forward).

Approximately nine in 10 headlines coded (89%) were traditional headlines that serve to summarize the story. Articles with clickbait headlines, where uncertainty is created by posing a question or leading the audience to read forward, had significantly fewer page views.[ref In two regression models predicting story page views and social referrals (with controls for the news outlet, article word count, type of article, campaign race type, strategy mentions, and issue mentions), clickbait headlines significantly reduce the number of page views (B=-.57 SE=.22; p=.01), but have no influence on social referrals (B=-.37 SE=.31; p=.24). The dependent variables – page views and social referrals – were modeled using negative binomial regression. In an OLS regression model predicting average time on page with a news story (with controls for the news outlet, article word count, type of article, campaign race type, strategy mentions, and issue mentions), headline type had no effect on read time (B=.19 SE=.18; p=.30).]

As the graph below illustrates, articles with a clickbait headline (e.g. “Who will show up to vote on Election Day?”) saw 1,276 fewer predicted page views compared to articles with a traditional headline.

[chart slug=”electioncoverage-7″]

Fact-checking is rare in local election news coverage

Local news stories also were examined for whether they were structured as fact-check articles, which explicitly referenced and evaluated campaign claims. We also analyzed whether the articles had veracity checks embedded within a news story that judged whether campaign claims were true or false.

A very small percentage of stories (2%) were framed as an official fact-check. A similarly small percentage of articles (6%) included the journalist issuing a veracity check of a claim. These veracity references included the news article mentioning something that may have been mistakenly, falsely, or truthfully stated. Official fact-check articles contained a similar number of issue mentions and strategy components compared to non-fact-check stories.

Given the relative infrequency of both of these article characteristics, we combined fact-checks and veracity checks into a single measure and evaluated whether their inclusion related to social referrals, time on page, or page views. We found that fact-checking articles performed about equally well in engagement as the vast majority of non-fact-checking articles.