skip to main content


Title: Anticipating Attention: On the Predictability of News Headline Tests
Headlines play an important role in both news audiences' attention decisions online and in news organizations’ efforts to attract that attention. A large body of research focuses on developing generally applicable heuristics for more effective headline writing. In this work, we measure the importance of a number of theoretically motivated textual features to headline performance. Using a corpus of hundreds of thousands of headline A/B tests run by hundreds of news publishers, we develop and evaluate a machine-learned model to predict headline testing outcomes. We find that the model exhibits modest performance above baseline and further estimate an empirical upper bound for such content-based prediction in this domain, indicating an important role for non-content-based factors in test outcomes. Together, these results suggest that any particular headline writing approach has only a marginal impact, and that understanding reader behavior and headline context are key to predicting news attention decisions.  more » « less
Award ID(s):
1717330
NSF-PAR ID:
10301845
Author(s) / Creator(s):
; ;
Date Published:
Journal Name:
Digital Journalism
ISSN:
2167-0811
Page Range / eLocation ID:
1 to 22
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. The culture within engineering colleges and departments has been historically quiet when considering social justice issues. Often the faculty in those departments are less concerned with social issues and are primarily focused on their disciplines and the concrete ways that they can make impacts academically and professionally in their respective arena’s. However, with the social climate of the United States shifting ever more towards a politically charged climate, and current events, particularly the protests against police brutality in recent years, faculty and students are constantly inundated with news of injustices happening in our society. The murder of George Floyd on May 25th 2020 sent shockwaves across the United States and the world. The video captured of his death shared across the globe brought everyone’s attention to the glaringly ugly problem of police brutality, paired with the COVID-19 pandemic, and US election year, the conditions were just right for a social activist movement to grow to a size that no one could ignore. Emmanuel Acho spoke out, motivated by injustices seen in the George Floyd murder, initially with podcasts and then by writing his book “Uncomfortable Converstations with a Black Man” [1]. In his book he touched on various social justice issues such as: racial terminology (i.e., Black or African American), implicit biases, white privilege, cultural appropriation, stereotypes (e.g., the “angry black man”), racial slurs (particularly the n-word), systemic racism, the myth of reverse racism, the criminal justice system, the struggles faced by black families, interracial families, allyship, and anti-racism. Students and faculty at Anonymous University felt compelled to set aside the time to meet and discuss this book in depth through the video conferencing client Zoom. In these meetings diverse facilitators were tasked with bringing the topics discussed by Acho in his book into conversation and pushing attendees of these meetings to consider those topics critically and personally. In an effort to avoid tasking attendees with reading homework to be able to participate in these discussions, the discussed chapter of the audiobook version of Acho’s book was played at the beginning of each meeting. Each audiobook chapter lasted between fifteen and twenty minutes, after which forty to forty-five minutes were left in the hour-long meetings to discuss the content of the chapter in question. Efforts by students and faculty were made to examine how some of the teachings of the book could be implemented into their lives and at Anonymous University. For broader topics, they would relate the content back to their personal lives (e.g., raising their children to be anti-racist and their experiences with racism in American and international cultures). Each meeting was recorded for posterity in the event that those conversations would be used in a paper such as this. Each meeting had at least one facilitator whose main role was to provide discussion prompts based on the chapter and ensure that the meeting environment was safe and inclusive. Naturally, some chapters address topics that are highly personal to some participants, so it was vital that all participants felt comfortable and supported to share their thoughts and experiences. The facilitator would intervene if the conversation veered in an aggressive direction. For example, if a participant starts an argument with another participant in a non-constructive manner, e.g., arguing over the definition of ethnicity, then the facilitator will interrupt, clear the air to bring the group back to a common ground, and then continue the discussion. Otherwise, participants were allowed to steer the direction of the conversation as new avenues of discussion popped up. These meetings were recorded with the goal of returning to these conversations and analyzing the conversations between attendees. Grounded theory will be used to first assess the most prominent themes of discussion between attendees for each meeting [2]. Attendees will be contacted to expressly ask their permission to have their words and thoughts used in this work, and upon agreement that data will begin to be processed. Select attendees will be asked to participate in focus group discussions, which will also be recorded via Zoom. These discussions will focus around the themes pulled from general discussion and will aim to dive deeper into the impact that this experience has had on them as either students or faculty members. A set of questions will be developed as prompts, but conversation is expected to evolve organically as these focus groups interact. These sessions will be scheduled for an hour, and a set of four focus groups with four participants are expected to participate for a total of sixteen total focus group participants. We hope to uncover how this experience changed the lives of the participants and present a model of how conversations such as this can promote diversity, equity, inclusion, and access activities amongst faculty and students outside of formal programs and strategic plans that are implemented at university, college, or departmental levels. 
    more » « less
  2. Sparked by a collaboration between academic researchers and science media professionals, this study sought to test three commonly used headline formats that vary based on whether (and, if so, how) important information is left out of a headline to encourage participants to read the corresponding article; these formats are traditionally-formatted headlines, forward-referencing headlines, and question-based headlines. Although headline format did not influence story selection or engagement, it did influence participants evaluations of both the headline’s and the story’s credibility (question-based headlines were viewed as the least credible). Moreover, individuals’ science curiosity and political views predicted their engagement with environmental stories as well as their views about the credibility of the headline and story. Thus, headline formats appear to play a significant role in audience’s perceptions of online news stories, and science news professionals ought to consider the effects different formats have on readers. 
    more » « less
  3. Audience analytics are an increasingly essential part of the modern newsroom as publishers seek to maximize the reach and commercial potential of their content. On top of a wealth of audience data collected, algorithmic approaches can then be applied with an eye towards predicting and optimizing the performance of content based on historical patterns. This work focuses specifically on content optimization practices surrounding the use of A/B headline testing in newsrooms. Using such approaches, digital newsrooms might audience-test as many as a dozen headlines per article, collecting data that allows an optimization algorithm to converge on the headline that is best with respect to some metric, such as the click-through rate. This article presents the results of an interview study which illuminate the ways in which A/B testing algorithms are changing workflow and headline writing practices, as well as the social dynamics shaping this process and its implementation within US newsrooms. 
    more » « less
  4. The spread of misinformation is a pressing societal challenge. Prior work shows that shifting attention to accuracy increases the quality of people’s news-sharing decisions. However, researchers disagree on whether accuracy-prompt interventions work for U.S. Republicans/conservatives and whether partisanship moderates the effect. In this preregistered adversarial collaboration, we tested this question using a multiverse meta-analysis ( k = 21; N = 27,828). In all 70 models, accuracy prompts improved sharing discernment among Republicans/conservatives. We observed significant partisan moderation for single-headline “evaluation” treatments (a critical test for one research team) such that the effect was stronger among Democrats than Republicans. However, this moderation was not consistently robust across different operationalizations of ideology/partisanship, exclusion criteria, or treatment type. Overall, we observed significant partisan moderation in 50% of specifications (all of which were considered critical for the other team). We discuss the conditions under which moderation is observed and offer interpretations.

     
    more » « less
  5. We aim to develop methods for understanding how multimedia news exposure can affect people’s emotional responses, and we especially focus on news content related to gun violence, a very important yet polarizing issue in the U.S. We created the dataset NEmo+ by significantly extending the U.S. gun violence news-to-emotions dataset, BU-NEmo, from 320 to 1,297 news headline and lead image pairings and collecting 38,910 annotations in a large crowdsourcing experiment. In curating the NEmo+ dataset, we developed methods to identify news items that will trigger similar versus divergent emotional responses. For news items that trigger similar emotional responses, we compiled them into the NEmo+-Consensus dataset. We benchmark models on this dataset that predict a person’s dominant emotional response toward the target news item (single-label prediction). On the full NEmo+ dataset, containing news items that would lead to both differing and similar emotional responses, we also benchmark models for the novel task of predicting the distribution of evoked emotional responses in humans when presented with multi-modal news content. Our single-label and multi-label prediction models outperform baselines by large margins across several metrics. 
    more » « less