Visual Propaganda in Chinese Central and Local News Agencies: A Douyin Case Study

The study probes how Chinese central and local news agencies employ visual propaganda on Douyin, amid governments’ growing use of social media and visuals to mobilize public emotions. Against the backdrop of China’s extensive governmental presence on short-video platforms, the paper argues that visual content is central to persuasion and emotion elicitation. Prior research has explored Chinese propaganda and visual strategies but has rarely compared central versus local governmental utilization, nor has it focused deeply on short videos due to their multimodal complexity. This study addresses that gap. The aims are twofold: (1) to compare the use of short videos at two levels of political subjectivity—central (directly under the CCP’s Propaganda Department) and local (provincial/ministerial)—and (2) to develop and test an analytical framework for visual propaganda that integrates visual and linguistic context. The research questions are: (1) What visual themes are constructed by central and local governments on Douyin? (2) Which kinds of characters are highlighted? (3) Under visual rhetorical strategies, which emotional responses are mainly stimulated by character images under different themes?

The paper reviews classic and contemporary scholarship on propaganda as deliberate, systematic attempts to shape perceptions and behaviors, with mass media as a key conduit. With the rise of digital media and short videos, political communication has become cheaper, faster, and more participatory, but also more manipulable. In China, propaganda encompasses hard forms (e.g., overtly pro-regime or anti-adversary news) and soft forms (entertainment-oriented, emotionally mobilizing content). Within China’s central–local political matrix (Tiao/Kuai), constrained decentralization shapes how local actors translate and localize central directives. Chinese propaganda is marked by nationalism and emotional governance, mobilizing sentiments such as fear, anger, pride, and respect. Visual narratives and imagery are especially potent in eliciting emotion and shaping identity; short videos, aligned with social media logic, are effective vehicles for both nationalist and everyday-life storytelling. Existing work often treats Chinese media as monolithic; this study differentiates central and local media to examine divergences in themes, visual characters, and emotional strategies.

The study adopts an integrative mixed-methods approach combining visual and textual analysis, pairing quantitative content coding with qualitative interpretation. Sampling: The two most influential central accounts (People’s Daily, CCTV News) and two leading local accounts (Sichuan Observation, The Paper) on Douyin were selected. Timeframe: March 29, 2018–July 28, 2023. Inclusion criterion: videos with more than 2 million likes to capture high-impact propaganda. Manual review ensured data quality; duplicates were removed. Final sample: 2,852 videos (central: 2,598; local: 254). Descriptive platform metrics (as of reporting) indicated substantial follower bases and total likes across accounts. Coding scheme: Guided by visual grammar/narrative theory, propaganda, and emotion research. Unit of analysis: individual short videos. Three dimensions were coded: (1) Themes (five categories): Military/police/firefighting; International diplomacy; Pandemic prevention and control (COVID-19); People’s livelihood and warmth; Festivals and anniversaries. (2) Visual characters (five categories): Soldiers/police/firefighters; Political leaders; Medical staff; Other influencers with no obvious political affiliation; Null lens (no main character). (3) Emotional propaganda (four categories based on Plutchik/Izard-informed labels): Anger/disgust/fear/intolerance; Joy/calm/pleasantness; Surprise/shock/confusion; Anticipation/acceptance/respect. Reliability and statistics: 10% of videos (n=200) were double-coded; Scott’s pi=0.881 (average reliability of items=0.878). Internal consistency checks (Cronbach’s alpha) exceeded 0.7 across scales; KMO values >0.7 indicated suitable correlations. Group differences were tested via chi-square. SPSS was used for reliability analyses. Selected case videos were qualitatively analyzed to illustrate governance performance and communication effects.

• Posting output and popularity: Since 2018, the accounts posted 5,325 (People’s Daily), 7,720 (CCTV News), 22,000 (Sichuan Observation), and 27,000 (The Paper) videos; videos exceeding 2 million likes totaled 1,619; 979; 235; and 19, respectively (total analyzed: 2,852).
• Themes (RQ1): Central vs local distributions (Table 3) differ significantly (χ²=16.66, p<0.001). Central (n=2,598): Military/police/firefighting 27.6% (719); International diplomacy 22.1% (572); People’s livelihood and warmth 20.6% (534); Pandemic prevention and control 17.4% (452); Festivals/anniversaries 12.3% (321). Local (n=254): People’s livelihood and warmth 70.9% (180); Festivals/anniversaries 14.6% (37); Military/police/firefighting 7.0% (18); International diplomacy 3.9% (10); Pandemic 3.6% (9).
• Visual characters (RQ2): Distributions differ significantly (χ²=40.01, p<0.001; Table 4). Central: Soldiers/police/firefighters 31.5% (815); Political leaders 25.9% (672); Other influencers 15.9% (414); Null lens 14.8% (386); Medical staff 11.9% (311). Local: Other influencers/no political affiliation 62.9% (160); Null lens 22.8% (58); Soldiers/police/firefighters 7.3% (18); Political leaders 3.9% (10); Medical staff 3.1% (8).
• Emotional propaganda (RQ3): Distributions differ significantly (χ²=57.39, p<0.001; Table 5). Central: Anger/disgust/fear/intolerance 39.7% (1,033); Anticipation/acceptance/respect 30.9% (803); Joy/calm/pleasantness 21.2% (552); Surprise/shock/confusion 8.2% (210). Local: Anticipation/acceptance/respect 58% (147); Joy/calm/pleasantness 27.5% (70); Anger/disgust/fear/intolerance 10.6% (27); Surprise/shock/confusion 3.9% (10).
• Emotion by theme (Table 6): Military/police/firefighting primarily evokes Anticipation/acceptance/respect (58.2%). International diplomacy and Pandemic prevention/control predominantly use Anger/disgust/fear/intolerance (68.2% and 53.1%). People’s livelihood and warmth and Festivals/anniversaries emphasize Joy/calm/pleasantness (45.7% and 41.6%).
• Emotion by visual characters (Table 7): Political leaders and Null lens are mostly labeled Anger/disgust/fear/intolerance (76.3% and 64.6%). Soldiers/police/firefighters and Medical staff mostly elicit Anticipation/acceptance/respect (57.2% and 65.5%). Influencers/no political affiliation are most often Joy/calm/pleasantness (57.7%).
• Exemplary most-liked videos: People’s Daily (34.967 million likes, military training); CCTV News (34.599 million likes, mourning scientist Yuan Longping—livelihood/warmth); Sichuan Observation (11.356 million likes, Chang Bai Mountain landscape with Buddha’s light—livelihood/warmth); The Paper (approx. 17.403 million likes, farewell ceremony for Yuan Longping—livelihood/warmth). Overall: Central agencies prioritize military/diplomacy and use more negative-valence mobilization (anger/fear), while local agencies focus on people’s livelihood and warmth, using anticipatory and respectful positive emotions.

Findings address all three RQs: (1) Central agencies emphasize national power and diplomacy; local agencies emphasize everyday life and festive rituals. (2) Central accounts foreground state apparatus and leadership, whereas local accounts feature nonpolitical figures and scenery. (3) Emotional strategies align with these choices: central agencies rely more on negative-valence mobilization around international/domestic crises, consistent with nationalist framing and emotional governance; local agencies prioritize positive sentiments (anticipation, acceptance, respect; joy) to cultivate affinity with local governance and public service images. The results reflect China’s hierarchical propaganda matrix. Central media sustain national narratives and political stability using hard propaganda elements and negative-valence emotions, particularly in diplomacy and crisis. Local media operate with greater stylistic latitude but within thematic constraints, adopting soft propaganda that packages ideology in entertaining, nonpolitical forms to build trust, warmth, and legitimacy. This differentiation illustrates constrained decentralization: local outlets can innovate around “people’s livelihood and warmth,” yet are limited in covering sensitive, unpredictable issues (e.g., pandemics, crises). The interplay between social media logic and governance logic enables both softening (to reduce resistance) and hardening (to reinforce authority) of propaganda across administrative tiers. The observed trust and reach advantages of central outlets underscore their continued authority in agenda setting and emotional framing.

The study contributes an integrative framework for analyzing visual propaganda in short videos and empirically differentiates strategies across central and local Chinese news agencies on Douyin. It shows that central media stress military and diplomatic themes, state-aligned visual characters, and negative-valence mobilization to fortify national narratives; local media emphasize livelihood warmth, nonpolitical figures, and positive emotions to build local legitimacy and affinity. Together, these patterns reveal how hierarchical political authority shapes both content and emotion in the short-video era, combining hard and soft propaganda. Future research should extend beyond one platform, incorporate audience-level analyses to assess persuasion and behavioral effects, and examine additional contexts and visual genres to understand long-term impacts of visual politics on public opinion and social dynamics.

The sample is limited to a single short-video platform (Douyin), which may constrain generalizability to other platforms. The study does not include direct audience-effect measures; future work should conduct quantitative analyses of governmental short videos and audience responses to evaluate effectiveness.