ABSTRACT
We examined whether news media literacy (NML) messages attenuate selective exposure and avoidance. One week before the 2016 U.S. presidential election, participants were randomly assigned to see a NML video advertisement before entering a simulated news aggregation website where behaviors were unobtrusively tracked. For three of the four NML messages, higher levels of partisan selective exposure among Republicans compared to Democrats in the control condition were reduced to non-significance. There were no effects on selective avoidance for either group. Several NML messages limited partisan selective exposure among Republicans, offering a concrete option for addressing problems of selective exposure online.
Acknowledgments
We would like to thank Ethan Degross from the Digital Scholarship & Publishing Studio at the University of Iowa for supporting the project.
Disclosure Statement
No competing financial interests exist.
Notes
1. Independents in both control conditions (no video or Texting PSA) were asked to rate each of the eight stories on a seven-point scale from “Very conservative” (1) to “Very liberal” (7). Overall, Independents saw the liberal stories as somewhat liberal (M = 3.59, S.D. = .77), and the conservative stories as slightly conservative (M = 4.24, S.D. = .76), with one “conservative” story (the SNL skit mocking Hillary Clinton) not seen as conservative (M = 3.80, S.D. = 1.06). We maintain our categorization of the videos, but discuss this limitation of our design.
2. We excluded participants from a sixth experimental condition (N = 68), where participants did not view a PSA before entering the news aggregator website because they selected on average an additional half a news story compared to the other conditions (t = 2.34, p = .02, M (no video) = 2.09, M (video) = 1.52), limiting comparability.
3. The final sample is similar in age (M = 35.95, S.D. = 11.38), education (M = 4.07, S.D. = 1.23), and gender (58.4% female) to the complete sample.
4. Among Democrats, the pairwise comparisons are all insignificant, p > .19.
5. Please contact the lead author for additional details on these analyses.