Motivation, Attention, and Visual Platform Design: How Moral Contagions Spread on TikTok and Instagram in the 2024 United States Presidential Election

1. Problem Statement

The study addresses a critical gap in political communication research: how moralization operates differently across visual-first social media platforms. While existing literature establishes that "moral contagions" (content framing issues as right/wrong) spread faster than neutral information, it remains unclear how platform-specific architectures (algorithms, network structures) and demographics influence which issues become moralized and how they spread.

Specifically, the authors investigate whether visual platforms uniformly accelerate moralization or if they create distinct "moral-political ecosystems." They focus on three central 2024 election issues: abortion (traditionally moralized), immigration (traditionally moralized), and the economy (traditionally pragmatic). The core puzzle is whether the economy, typically a pragmatic topic, becomes moralized differently on TikTok versus Instagram.

2. Methodology

Data Collection

• Dataset: A massive corpus of 2,027,595 TikToks (Jan 1 – Nov 14, 2024) and 1,126,972 Instagram posts (Jan 1 – Oct 28, 2024).
• Sources: Meta Content Library (Instagram) and the official TikTok Researcher API.
• Sampling: Snowball sampling based on a curated list of keywords and hashtags to filter for political relevance.

Content Categorization

Posts were classified into four categories based on hashtag usage:

1. Pro-life Abortion
2. Pro-choice Abortion
3. Immigration (Right-leaning focus)
4. Economy (Politically neutral focus, later expanded to include cryptocurrency)

Quantitative Metrics

• Supply: Total volume of posts per category per day.
• Demand: Average views per post per day (measuring attention/virality).
• Volatility: Statistical variance in supply and demand over time, particularly before and after Kamala Harris's nomination.

Moral Foundations Analysis (eMFD)

• Tool: The extended Moral Foundations Dictionary (eMFD) was applied to post captions.
• Mechanism: Unlike binary dictionaries, eMFD uses probabilistic word-foundation associations to score the intensity of five moral foundations: Care, Fairness, Loyalty, Authority, and Sanctity.
• Sentiment Weighting: Scores were adjusted for emotional valence (positive/negative sentiment) to distinguish between virtue (e.g., compassion) and vice (e.g., harm).

Network Analysis

• Technique: Semantic network analysis of hashtag co-occurrence.
• Visualization: Top 1,000 hashtag pairs were mapped using NetworkX and Netwulf to visualize the topology of political discourse (connectivity vs. fragmentation).

3. Key Contributions

1. Platform-Specific Moral Ecosystems: The study demonstrates that moralization is not an inherent property of an issue but a product of the interaction between audience demographics, platform architecture, and partisan framing.
2. Re-moralization of Pragmatic Issues: It provides empirical evidence that "pragmatic" economic issues (specifically cryptocurrency regulation) can be rapidly moralized through binding foundations (Loyalty and Authority), framing regulation as government overreach.
3. Algorithmic Divergence in Supply/Demand: The paper reveals a fundamental decoupling of supply and demand on TikTok versus their alignment on Instagram, challenging the assumption that visual platforms uniformly amplify moral content.
4. Structural Impact on Candidate Positioning: It links network topology (circular vs. fragmented) to the political viability of candidates, showing how platform design can structurally isolate candidates (e.g., Kamala Harris) from key discourse clusters (e.g., the economy).

4. Key Results

A. Supply and Demand Dynamics (Attention)

• TikTok (Content-Driven):
  • Supply: Dominated by Economy content (37,700 posts).
  • Demand: Dominated by Moralized content (Pro-life and Immigration).
  • Dynamics: High volatility. After Harris's nomination, supply volatility dropped by 96%, shifting from sporadic bursts to sustained production.
  • Virality: "Dark horse" effect: Low-supply moral topics achieved massive viral peaks (up to 105k views/post).
• Instagram (Network-Driven):
  • Supply & Demand: Aligned. Economy content led both in volume and views (avg. ~10k views/post).
  • Dynamics: Stable, low volatility. Harris's nomination had no significant stabilizing effect.
  • Virality: Lacked extreme outliers; high performance was consistent for pragmatic topics.

B. Moral Foundations and Framing (Motivation)

• Traditional Issues: Abortion and Immigration consistently invoked negative moral framing (vice-related) across both platforms.
• The Economy & Cryptocurrency:
  • TikTok: Economy content showed negative Care but positive Loyalty.
  • Instagram: Economy content showed strong positive Loyalty (3.96) and Authority (3.72).
  • Crypto Specifics: Cryptocurrency discourse achieved the highest Loyalty scores of any category (TikTok: 5.36, Instagram: 3.46), framing financial freedom as an in-group identity and regulation as "subversion" of authority.
• Candidate Framing:
  • Trump: Consistently less negative framing.
  • Harris: Positioned centrally on Loyalty/Authority but more negative on Care.
  • Biden: Strongly connected to economy content (often negatively) and crypto criticism.

C. Network Topology (Design)

• TikTok (Circular Topology): Exhibits a highly interconnected, circular structure. Hashtags from different ideological camps (e.g., pro-life, pro-choice, economy) share intermediary nodes, facilitating cross-cutting exposure.
• Instagram (Fragmented Topology): Shows distinct, semi-isolated clusters with weak bridges.
  • Harris's Isolation: The #KamalaHarris node clustered tightly with Pro-choice content but was physically distant from the Economy cluster. This structural isolation reinforced her exclusion from economic discourse.
  • Crypto Clustering: On TikTok, crypto clustered with Biden (criticism of regulation); on Instagram, it clustered with Trump (supportive framing).

5. Significance and Implications

• Theoretical Advancement: The study validates and extends the MAD (Motivation, Attention, Design) model. It proves that platform design (algorithms) fundamentally moderates moral contagion. TikTok's algorithm allows moral content to overcome supply disadvantages, while Instagram's network model amplifies established accounts discussing pragmatic issues.
• Political Strategy: The findings suggest that "moralization" is a strategic tool that can be applied to non-moral issues (like the economy) depending on the platform. Campaigns must tailor framing (e.g., using Loyalty/Authority for crypto on Instagram) to specific platform ecosystems.
• Democratic Discourse: The structural fragmentation on Instagram may reinforce polarization by isolating candidates from critical issue clusters (e.g., Harris from the economy). Conversely, TikTok's circular topology allows for broader, albeit potentially chaotic, cross-ideological exposure.
• Methodological Shift: The paper highlights the necessity of moving beyond text-only analysis in the era of visual media, suggesting future work must integrate computer vision and audio analysis to capture moral signals not present in captions.

In conclusion, the paper argues that understanding political moralization requires examining the platform-specific ecosystem where architecture, demographics, and content strategy interact to determine which issues get moralized and how they spread.