Selective exposure and selective avoidance of algorithm-written news stories

New study “ Source Credibility Matters: Does Automated Journalism Inspire Selective Exposure?” by Chenyan Jia and Thomas J. Johnson from University of Texas in Austin investigated an emerging field of journalism: automated journalism. 

Specifically, how selective exposure and selective avoidance of news stories was practiced comparing stories that were declared to have been written by algorithms vs. stories declared to have been written by human authors.

The topics chosen for the experiment were contentious: gun control and abortion. The stories were in three categories by attitude: attitude-consistent news, attitude-challenging news, and neutral. The study was online and there were 351 participants.

A growing number of newsrooms employ algorithmic generation of news particularly in fact-based topics such as sports and financial news. However, many argue that automated news can be applied to other domains. 

Previous research on automated news has revealed that people tend to view automated news as less biased than human-generated news. On the topic of selective exposure, previous studies suggest that people tend to avoid attitude-challenging news, but some actively seek out such. Since people also prefer credible information, they tend to view attitude-consistent news as more credible.

The study at hand found out that for both human written news and algorithmic news, the results were in line with the previous studies and the participants preferred attitude-consistent news and rated them as more credible. 

Contrary to theorizations that algorithmic news might be a good way to gain credible exposure to attitude-challenging news, it was found out that the participants rated algorithm-generated attitude-challenging news also as less credible. Thus, it was found out that both selective exposure and avoidance occur alike when it comes to algorithmic news. 

Of the topics in question, the topic of guns differed from the others, as in it, the participants were more likely to select human-authored attitude-consistent news. The authors speculated that since the topic of guns is related to public security rather than, say, personal choice, source credibility was viewed as more important and human-generated news were viewed as more credible overall in the study.

The results have relevance for both the field of selective exposure, as source credibility was more affected by attitude than by author type, and for the emerging field of studies on automated journalism.

The article “ Source Credibility Matters: Does Automated Journalism Inspire Selective Exposure?” by Chenyan Jia and Thomas J. Johnson is in International Journal of Communication. (free access).

Picture: Untitled by Brett Jordan.

License Unsplash.

Give us feedback