Framing the Fray: Evaluating Conflict Frames in Indian Election News Coverage

TL;DR

This study scrutinizes the use of conflict frames in online English-language Indian election coverage across seven outlets during the 2014 and 2019 general elections. It develops a DistilBERT-based frame classifier trained on headline data, links frames to outlet modality and party type, and compares coverage to campaign speeches to assess under- and over-reporting of issues. Key findings show TV-based outlets exhibit more conflict framing than print outlets, ideological bias is not the primary driver, and national parties are more often embedded in conflict frames; moreover, news tends to emphasize attacks on the opposition while under-reporting substantive electoral issues discussed in speeches. The work highlights potential implications for polarization and political knowledge and provides a publicly available dataset and analysis pipeline for replication and extension.

Abstract

In covering elections, journalists often use conflict frames which depict events and issues as adversarial, often highlighting confrontations between opposing parties. Although conflict frames result in more citizen engagement, they may distract from substantive policy discussion. In this work, we analyze the use of conflict frames in online English-language news articles by seven major news outlets in the 2014 and 2019 Indian general elections. We find that the use of conflict frames is not linked to the news outlets' ideological biases but is associated with TV-based (rather than print-based) media. Further, the majority of news outlets do not exhibit ideological biases in portraying parties as aggressors or targets in articles with conflict frames. Finally, comparing news articles reporting on political speeches to their original speech transcripts, we find that, on average, news outlets tend to consistently report on attacks on the opposition party in the speeches but under-report on more substantive electoral issues covered in the speeches such as farmers' issues and infrastructure.

Figures (4)

• Figure 1: 95% CI of probabilities for major political coalitions as aggressors (a) and targets (b) in conflict frame headlines across media outlets. Left-leaning media and UPA are shown in blue; right-leaning media and NDA are in orange. The CI values and estimates are detailed in Table \ref{['tab:alliance_attacked']} and Table \ref{['tab:alliance_being_attacked']}.
• Figure 2: The frequency of speeches mentioning a specific issue versus the proportion of articles mentioning the same issue. Analysis shows substantive topics, like farmers' issues, are underrepresented (bottom-left), whereas the most frequently mentioned issue is the opposition. The frequency and proportion values are detailed in Table \ref{['tab:issues_table']}.
• Figure 3: An example of how parties are extracted for a particular article.
• Figure 11: Proportion of headlines with coalition as aggressor