Framing the Crisis · Signature Work

This was my undergraduate Signature Work at Duke Kunshan University, submitted on March 7, 2024 under Prof. Andrew Macdonald in the Division of Social Science. My roommate, Xingyu Shen, helped collect and clean the data.

Title: Framing the Crisis: A Study of Local vs. Central Media Propaganda During Shanghai COVID-19 Lockdown.

The question

When central and local state media cover the same crisis in the same language, how do their propaganda strategies differ, and is the relationship one of conflict or of coordination?

The corpus

• 1,161 articles across the April 1 to June 1, 2022 lockdown window
• Central: People's Daily (207) + Global Times (179), giving 386 articles
• Local: Jiefang Daily 解放日报 (316) + Xinmin Evening Post 新民晚报 (459), giving 775 articles
• A Python crawler against CNKI for People's Daily, Global Times, and Jiefang Daily, plus the Xinmin website; Python translation pass for analysis

The method

I used two methods that could check and deepen each other:

1. LDA topic modeling (Latent Dirichlet Allocation) across the full corpus. Topic coherence pointed to nine topics for central media and ten for local media; I then named each topic from its twenty highest-weight words.
2. Hand-coded content analysis of a stratified sample of 76 articles (9 central and 67 local), drawn from the beginning, midpoint, and end of the lockdown. I coded source (officials / professionals / ordinary people), article type (information / persuasion / narrative), tone (authoritative / suggestive / warm / sympathetic), and frame (struggle / resilience / government response / local response / economy / health).

What I found

The dominant relationship was not conflict. It was a layered division of labor: two parts of one apparatus performing complementary tasks.

• Central media ran what I'd call hard propaganda. Authoritative tone (56% against local's 24%). Government officials as the dominant source (67% against 25%). Information-sharing as the dominant nature (78% against 34%). Themes built around national cohesion, governmental capacity, and global leadership.
• Local media ran soft propaganda. Warm and sympathetic tone (61% warm against central's 11%). Ordinary people as the dominant source (51% against 11%). Narrative storytelling as the dominant form (55% against 11%). Themes built around community resilience, neighborhood logistics, and human-centric stories.

The most striking finding unfolded over time. In the hand-coded local-media sample from May 2022, at the worst of the lockdown, 91% of articles used a resilience frame. As public morale frayed, local coverage narrowed toward endurance, mutual help, and holding on.

One system, two jobs

An authoritarian propaganda system is not a single block. It has levels and a division of labor: central outlets signal and control information, while local outlets perform empathy and manage morale. Whether they agree is not the interesting question; of course they agree. The question is how the work of sustaining that agreement gets divided.

The portfolio page summarizes the method and results. The underlying newspaper corpus is not redistributed here because the source articles remain copyrighted.