The processing and evaluation of news content on social media is influenced by peer-user commentary

The proliferation of social media has fundamentally altered information dissemination, supplementing news content with various social cues like comments and Likes. This study examines the impact of these cues on how users process and evaluate online news. Prior research on conformity and the bandwagon effect (Asch, 1951; Nadeau et al., 1993; Sundar, 2008) suggests that individuals adjust their opinions to align with perceived majority views. Likes, as paralinguistic digital affordances (PDAs), and comments can act as this "bandwagon," influencing observers (Bak & Keßler, 2012; Sherman et al., 2016). While previous studies have compared high and low numbers of Likes (Winter et al., 2015), this study uniquely investigates the impact of the presence versus absence of Likes, a practice increasingly adopted by social media platforms. The study hypothesizes that Likes will improve attitudes, sharing intent, agreement with content, perceived public attitude, and credibility (H1). Regarding comments, prior research suggests they influence perceptions and evaluations, with negative comments potentially being more persuasive than positive ones (Kim, 2018; Waddell, 2018a, 2018b; Winter et al., 2015). The study further explores the effect of comment type (positive, negative, mixed, none) on the same outcome variables (H2). Winter et al. (2015) found that negative comments were more persuasive than Likes and positive comments, potentially due to a ceiling effect of high agreement with the content. This study explores this idea, hypothesizing that Likes and positive comments will have a stronger influence when readers have a negative predisposition to the content (H3) and that negative comments will have a greater influence than positive comments regardless of predisposition (H4). The study uses three news articles—ideologically congruent, incongruent, and neutral—to examine the interaction between content type and the effects of Likes and comments, hypothesizing that Likes and comments will have a greater impact on ideologically neutral content (H5).

The study draws upon existing literature on conformity, the bandwagon effect, and the role of social cues in online environments. Asch's conformity experiments demonstrate the pressure individuals experience to conform to group opinions (Asch, 1951, 1955, 1956). The bandwagon effect describes the tendency of individuals to adopt beliefs and opinions prevalent in the larger population (Nadeau et al., 1993; Schmitt-Beck, 2015). This is often linked to the bandwagon heuristic, a cognitive bias where the majority opinion is perceived as correct (Sundar, 2008; Mutz, 1998). Prior research on social media has shown the impact of Likes (increasing likeability and likelihood of liking) (Bak & Keßler, 2012; Sherman et al., 2016) and comments (affecting acceptance and attitudes) (Kim, 2018; Waddell, 2018a, 2018b; Winter et al., 2015). Winter et al. (2015) examined the effects of Likes and comments on Facebook news stories, finding negative comments more influential than Likes or positive comments, potentially because of a ceiling effect caused by high levels of initial agreement with the article's content. The concept of negativity bias (Baumeister et al., 2001; Ito et al., 1998; Norris, 2021; Rozin & Royzman, 2001), where negative information has a stronger impact than positive information, is also relevant. Studies on anecdotal vs. scientific evidence in comments (Hinnant et al., 2016) also inform the study's design. The study also touches on the ecological validity of previous experiments, highlighting the need for more realistic interactive interfaces (Winter et al., 2015) and the influence of content type on reader predispositions.

A custom-built website mimicking Facebook was used to conduct an online experiment. The experiment employed a 2 x 4 x 3 mixed factorial design, manipulating Likes (present/absent—between-subjects), comments (positive, negative, mixed, none—between-subjects), and article content type (congruent, incongruent, neutral—within-subjects). Participants (412 undergraduate psychology students) were randomly assigned to one of eight conditions (Table 1). Three news articles were used (adapted from The Guardian): one about Greta Thunberg (congruent), one about video games and aggression (incongruent), and one about a tropical storm (neutral). The comments comprised subjective, emotive, and potentially fallacious rhetoric (Figure 1). After reading each article, participants rated five outcome variables (Table 2) using seven-point Likert scales: attitude, share intent, ideological congruence, perceived public attitude, and credibility. These variables were assessed using statements adapted from previous studies (Appelman & Sundar, 2016; Winter et al., 2015; Xu, 2013). A sequential analysis design (Lakens, 2014) with a pre-determined sample size and inference criteria was used. Mixed factorial ANOVAs were conducted to analyze the data, with Bonferroni corrections for multiple comparisons. The study was preregistered on Open Science Framework.

Five mixed factorial ANOVAs were performed, one for each outcome variable. Regarding attitude, a significant main effect of comment sentiment was found (Figure 2), with negative and mixed comments inducing more negative attitudes than positive comments. An interaction effect between article content type and comments was also observed (Figure 3), with comments having a stronger influence on attitudes towards neutral articles. Share intent showed a similar pattern, with negative and mixed comments significantly reducing the intent to share compared to positive comments (Figure 4). An interaction effect between content type and comments also emerged (Figure 5), with comments reducing the willingness to share the congruent article. Ideological congruence (agreement with article ideas) showed a significant main effect of comments (Figure 6), with negative and mixed comments reducing agreement. A significant main effect of article content type was also found. Perceived public attitude showed a significant main effect of comments (Figure 7) and a significant interaction between content type and comments, with comment sentiment having the greatest effect on neutral articles. Credibility was significantly affected by comment sentiment (Figure 8), with negative and mixed comments reducing perceived credibility. In all cases, Likes had no significant effect on any of the outcome variables. These findings support hypotheses H2 and H4, concerning the negative influence of negative comments. Hypothesis H3, concerning the interaction of negative predisposition and positive comments, was not fully supported but revealed a weak interaction, suggesting that in the absence of a strong predisposition, positive comments exert a greater influence. Hypothesis H5, concerning the greater influence on neutral content, was not supported.

The study's results demonstrate a strong influence of negative comments on the processing and evaluation of news articles on social media, even when comments contain subjective, emotive, and fallacious information. This finding highlights the power of negativity bias (Baumeister et al., 2001; Ito et al., 1998; Norris, 2021; Rozin & Royzman, 2001), where negative information captures attention and influences judgments more effectively than positive information. The unexpected lack of effect from Likes may be due to their less concrete nature compared to comments, aligning with exemplification theory (Zillmann, 1999) which suggests that concrete examples, such as comments, are more influential than aggregated data. The dual-processing theory (Barrett et al., 2004; Epstein, 1994) might also explain the differential impact, with comments engaging System 1 processing and Likes requiring more effortful System 2 processing. The finding that perceived public attitude was less susceptible to negativity bias suggests that while personal opinions are affected by negative comments, the perception of public opinion is not solely based on negative information. The possibility of authority bias (Milgram, 1963) may also explain the persuasiveness of negative comments, where readers perceive critics as more knowledgeable.

This study underscores the significant influence of negative peer-user comments on online news evaluations, particularly when these comments are subjective and potentially fallacious. The lack of a similar effect for Likes suggests different processing mechanisms for these social cues. These findings have important implications for news platforms considering comment sections and for social media users themselves who should critically evaluate online comments, considering the source's expertise and potential biases. Future research could explore age-related differences in negativity bias, investigate the role of social context in the processing of Likes, and examine the impact of comment characteristics on perceived authority.

The study's sample consisted primarily of young adult psychology students, potentially limiting the generalizability of findings to other demographics. The assumed ideological dispositions of participants regarding the articles did not fully align with expectations. Specifically, participants displayed less negativity towards the ideologically incongruent article than anticipated. This limitation, however, does not significantly undermine the key findings.