Avoid or Authenticate? A Multilevel Cross-Country Analysis of the Roles of Fake News Concern and News Fatigue on News Avoidance and Authentication

Abstract

Citizens these days feel inundated with news online and are worried about its veracity. This study examines if these concerns in the digital news environment led to greater news avoidance and news authentication behaviors. The relationships were tested across 16 countries by combining individual-level survey data from the Reuters Institute Digital News Report (N = 34,201) with country-level data based on comparative media systems research. Analysis from multilevel modeling showed that concern with fake news was related to news authentication and news fatigue was related to news avoidance. High news fatigue also accentuated the influence of concern with fake news on news avoidance while low fatigue attenuated the relationship. Additional cross-level interactions further contextualized the findings according to media system, showing how the relationships can vary under different conditions of press market, political parallelism, journalistic professionalism, and public service broadcasting. This study demonstrates the utility and importance of considering the contextual role of media system to understand individuals’ perceptions of news they receive online and subsequent news engagement, especially in the context of fake news research because its prevalence and deleterious impact varies across countries.

Keywords:

• Fake news
• news fatigue
• news avoidance
• news authentication
• media system
• multilevel analysis

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Notes

1 Although the Reuters dataset comprised 38 countries around the world, our sample selection was constrained by the availability of macro-level data from Brüggemann et al. (2014) that matched the Reuters data. Greece was featured in both Brüggemann et al. and the Reuter’s dataset, but certain questions required for this study were not asked in the Greek sample of the Reuter’s dataset.

2 In all moderation analysis we adopted values of the moderators based on the mean and ±1 standard deviation of their distributions as indicators of “low,” “middle,” and “high” levels.

3 The three-way interaction for journalistic professionalism was also significant (b = .02, p = .044). However, our application of the Benjamini-Hochberg (BH) procedure showed that the p value exceeded the BH critical value (.024), which suggested a possible false positive result.