The Impact of Exposure to Discordant Media on Political Polarization in Turkey

Researchers:
Ahmet Akbiyik
Jeremy Bowles
Shelley Liu
Fieldwork by:
Location:
Turkey
Sample:
3,851 individuals
Timeline:
2021
Initiative(s):
Governance Initiative
Target group:
  • Adults
Outcome of interest:
  • Attitudes and norms
Intervention type:
  • Media
AEA RCT registration number:
AEARCTR-0008489
Partners:

Political polarization can contribute to democratic backsliding and political violence. While previous research suggests that exposure to news media across the political spectrum may not reduce polarization in democracies, less is known about how media contributes to polarization in single-party states where pro-government media is the primary source of information for citizens. Researchers conducted a randomized evaluation to test the impact of exposure to and incentives to consume discordant media on political attitudes and behaviors in Turkey.

الموضوع الأساسي

Political polarization can contribute to democratic backsliding and political violence. As polarization increases, citizens may prefer to consume media from outlets that align with their partisan beliefs. In turn, consumers of media outlets displaying partisan bias may be more distrustful of people and media outlets with opposing viewpoints. As a media outlet’s audience becomes increasingly partisan, the outlet faces increasing incentives to produce content with the partisan slant that its consumers want.

Exposure to politically discordant or cross-partisan media (i.e., media that does not align with a citizen’s political beliefs) may influence citizens’ political beliefs, attitudes, and participation by exposing individuals to new information and framings of events. However, individuals may not accept information that contrasts with their preexisting beliefs, or individuals may prefer not to expose themselves to discordant media. While previous research suggests that exposure to news media across the political spectrum may not reduce polarization in democracies, less is known about how media contributes to polarization in single-party states, where pro-government media is the primary source of information for citizens.

سياق التقييم

During the twentieth century, democracy in Turkey was repeatedly interrupted by military coups.  Since 2002, the AKP (Justice and Development Party) has been the dominant political party, winning every parliamentary election up to 2022. Since then, international organizations have accused the Turkish government of subjecting journalists to harassment, physical violence, and imprisonment.1  In October 2022, new legislation gave the Turkish government the authority to imprison citizens and journalists for posting disinformation on social media.

Since 2002, the AKP has expanded its control of the media, as corporations with close ties to the AKP have acquired major media outlets and altered those outlets’ ideological stances to be more pro-government. Such changes in media ownership, along with newly emerging highly partisan news outlets, have contributed to the dominance of pro-government media in Turkey.

Adults were eligible to participate in the study if they lived in Turkey and actively used social media platforms. 70 percent of participants surveyed in 2021 read the news from pro-government outlets, while 46 percent were classified as having a pro-government political leaning. Sixty percent of those surveyed were male and 87 percent completed at least secondary education. The average age was 28.6 years.

Three women sitting and relaxing on stone ledge by water, looking at phones
Three women, sitting on a ledge near water, use their phones in Istanbul, Turkey.
Photo: Emilija Miljkovic | Shutterstock.com

معلومات تفصيلية عن التدخل

Researchers conducted a randomized evaluation to test the impact of exposure to and incentives to consume discordant media on political attitudes and behaviors. From February to August 2021, researchers recruited 3,851 adults. All participants downloaded an app onto their phones for data collection and implementation of the intervention. Once a week for seven months, three politically notable headlines from each outlet were flagged for participants as push notifications from the phone app. Based on each participant’s political beliefs, they were randomly assigned to one of five groups:  

  1. Strongly anti-government (886 participants): Participants in this group followed Gazete Duvar, an online news outlet with a strongly anti-government perspective. 
  2. Weakly anti-government (455 participants): Participants in this group followed Medyascope, an online video news outlet with a moderately anti-government perspective.
  3. Weakly pro-government (402 participants): Participants in this group followed Hürriyet, a newspaper with a moderately pro-government perspective. 
  4. Strongly pro-government (976 participants): Participants in this group followed Sabah, a newspaper with a strongly pro-government perspective.
  5. Comparison (1,132 participants): Participants in this group maintained their usual media consumption. 

Once a month, participants had the option to complete a quiz testing their memory of information in the previous month’s headlines. Participants received 10 TL ($0.31) for completing a quiz, and an additional 10 TL for correctly answering at least half of the questions. Participants in the comparison group received quizzes on unrelated topics.

Participants were never assigned to the group that aligned with their political beliefs (e.g., a person with strongly pro-government beliefs was never assigned to the strongly pro-government group). Some participants were assigned to groups that moderated their political beliefs (e.g., a person with strongly pro-government beliefs may have been assigned to the weakly pro-government group), but this study does not primarily focus on those participants.

Researchers surveyed participants four months after they enrolled on their engagement with discordant media, affinity for the AKP, attitudes on democracy, trust towards people with opposing political beliefs, and perceptions of bias in the media. After seven months, researchers surveyed participants again on the same topics.

النتائج والدروس المستفادة بشأن السياسات

Participants changed their media consumption habits, trusted discordant media sources more, and had less polarized attitudes towards the government, but did not change how much they trusted people with opposing political views. Results from a longer-term follow-up are forthcoming.

Discordant media exposure and consumption: Participants in all of the intervention groups were more likely to follow and see posts from media outlets with similar perspectives to the outlets they were assigned to. Moreover, based on data collected from participants who shared their Twitter accounts with researchers, 36 percent of participants continued following their assigned media outlet months after the intervention concluded. On an index measuring how many pro/anti-government outlets participants reported frequently seeking out and preferring to read, participants in the anti-government groups increased their consumption of anti-government news by 0.38 standard deviations, while participants in the pro-government groups increased pro-government media consumption by up to 0.14 standard deviations.

Moreover, participants in the weakly pro-government and strongly anti-government groups were less likely to click on anti-government and pro-government news, respectively. Participants in the anti-government groups were 0.16 to 0.22 standard deviations more aware of various anti-government news outlets by the end of the intervention. However, there was no impact on awareness of pro-government media for any of the intervention groups.

Political attitudes: By the end of the intervention, the anti- and pro-government groups report less and more affinity for the AKP, respectively. The weakly anti-government group reported being 10 percentage points (from a baseline of 32 percent) less likely to vote for the AKP, while the weakly pro-government group reported being 7 percentage points more likely to vote for the AKP. Participants in the strongly anti- and pro-government groups reported no change in their intentions to vote for the AKP. Meanwhile, the anti-government groups expressed more affinity for opposition parties, but their intended vote for the opposition did not change. The pro-government groups reported no change in affinity towards or intention to vote for opposition parties. 

During the intervention, the pro-government groups reported feeling more satisfied with democracy in Turkey, although this effect only persisted after the intervention for the strongly pro-government group. Participants in the strongly anti-government group were less favorable of the AKP’s performance on an index of policy issues during the intervention, but there was no effect on perceptions of AKP performance for either anti-government group after the intervention. Although the weakly anti-government group was more likely to express pro-democracy beliefs during the intervention, there was no effect on beliefs about democracy for any of the intervention groups after the intervention ended. 

Trust in media outlets and people with opposing views: Participants in the pro-government groups reported 0.11 to 0.20 standard deviations more trust in pro-government media relative to the comparison group, while those in the anti-government groups reported 0.18 to 0.25 standard deviations more trust in anti-government media at the end of the intervention. Researchers suggest that participants in the anti-government groups increased their trust in assigned outlets because they learned about a previously unknown source of information, while participants in the pro-government groups increased their trust in pro-government media because they were exposed to a greater breadth of information.

Participants also expressed more trust towards others with opposing political views during the intervention. However, the effects on trust disappeared by the end of the intervention. Researchers suggest that achieving lasting reductions in polarization may require more than just exposure to opposing views in media.

Overall, exposure to discordant media may be a potential strategy for mitigating political polarization, particularly when targeted at less polarized individuals. Future research could determine if discordant media exposure can have a sustained impact on individual attitudes.

The implementer used the research to inform its funding of independent media outlets in Turkey. In particular, based on the finding that the increase in trust in pro-government media appears to be explained by such outlets having broader coverage, the implementer plans to fund independent media to also produce content beyond political news to increase their perceived trustworthiness.

Akbiyik, Ahmet, Jeremy Bowles, Horacio Larreguy, and Shelley Liu. “Polarization and Exposure to Counter-Attitudinal Media in a Nondemocracy.” Working paper. November 2, 2023.

1.
Schenkkan, Nate, Susan Corke, Andrew Finkel, David J. Kramer, and Carla Robbins. “Corruption, Media, and Power in Turkey.” Freedom House, 2014. https://freedomhouse.org/report/special-report/2014/corruption-media-and-power-turkey