Examining the Role of YouTube Production and Consumption Dynamics on the Formation of Extreme Ideologies

Sarmad Chandio; Rishab Nithyanand

cs.CLMar 9, 2026

Examining the Role of YouTube Production and Consumption Dynamics on the Formation of Extreme Ideologies

Authors: Sarmad Chandio, Rishab Nithyanand

Organizations: University of Iowa

Abstract

The relationship between content production and consumption on algorithm-driven platforms like YouTube plays a critical role in shaping ideological behaviors. While prior work has largely focused on user behavior and algorithmic recommendations, the interplay between what is produced and what gets consumed, and its role in ideological shifts remains understudied. In this paper, we present a longitudinal, mixed-methods analysis combining one year of YouTube watch history with two waves of ideological surveys from 1,100 U.S. participants. We identify users who exhibited significant shifts toward more extreme ideologies and compare their content consumption and the production patterns of YouTube channels they engaged with to ideologically stable users. Our findings show that users who became more extreme consumed have different consumption habits from those who do not. This gets amplified by the fact that channels favored by users with extreme ideologies also have a higher affinity to produce content with a higher anger, grievance and other such markers. Lastly, using time series analysis, we examine whether content producers are the primary drivers of consumption behavior or merely responding to user demand.

Figures & tables

Figure 1: Distribution of maximum shifts in ideological attitudes and beliefs between the two waves.

Figure 2: This figure shows the temporal evolution of daily averaged anger scores for $\mathcal{G}_{ext}$ and $\mathcal{G}_{neu}$ over the year. The blue lines divide the timeline into four distinct quartiles. Read Section 9.3 in the Appendix for more detail on how the topic summaries for each quartile were computed.

Marker	$\Delta_{\text{across}}$	$\Delta_{\text{within}}^{\text{neu}}$			$\Delta_{\text{within}}^{\text{ext}}$
		Q2-Q1	Q3-Q2	Q4-Q3	Q2-Q1	Q3-Q2	Q4-Q3
Anger	6.8 *	0.9 *	20.5 *	–8.5 *	–17.0 *	5.9 *	–17.3 *
Griev.	13.0 *	–4.1 *	0.0	5.4 *	–12.5 *	-17 *	–13.1 *
Power	0.6 *	5.4 *	12.7 *	–14.3 *	–6.1 *	2.0 *	5.5 *
Neg.	1.8 *	-2.1 *	1.3 *	-0.2 *	4.5 *	0.4 *	0.2 *
In-group	0.0	2.7 *	–14.8 *	–2.8 *	–25.3 *	–3.7 *	–11.3 *

Table 1: Percentage differences in exposure to linguistic markers.

\Delta_{\text{across}}

shows the difference between

\mathcal{G}_{ext}

and

\mathcal{G}_{neu}

users over the year (

\mathcal{G}_{ext}

\mathcal{G}_{neu}

\Delta_{\text{within}}

shows the marginal percentage differences within

\mathcal{G}_{neu}

and

\mathcal{G}_{ext}

groups from one quarter to the next. Asterisks denote

p<0.05

Figure 3: This figure shows the temporal evolution of daily averaged anger scores for $\mathcal{C}_{high}$ and $\mathcal{C}_{low}$ over the year. The blue lines divide the timeline into four distinct quartiles. Read Section 9.3 in the Appendix for more detail on how the topic summaries for each quartile were computed.

Marker	$\Delta_{\text{across}}$	$\Delta_{\text{within}}^{\text{C-low}}$			$\Delta_{\text{within}}^{\text{C-high}}$
		Q2-Q1	Q3-Q2	Q4-Q3	Q2-Q1	Q3-Q2	Q4-Q3
Anger	39.8 *	–4.5	49.1 *	–18.2 *	–7.6	70.9 *	–14.0 *
Griev.	6.5 *	2.4	3.3	0.01	0.6	3.1	–0.8
Power	13.7 *	4.5	–0.9	2.9	7.6 *	3.4	2.7
Neg.	7.5 *	3.7 *	0.1	1.0	-1.4 *	4.1 *	-0.1
In-group	9.5	7.3	2.8	–35.4	–1.8	5.2	–7.6

Table 2: Percentage differences in production of content with markers of extremism.

\Delta_{\text{across}}

shows the difference between

\mathcal{C}_{high}

and

\mathcal{C}_{low}

channels over the year (

\mathcal{C}_{high}

\mathcal{C}_{low}

\Delta_{\text{within}}

shows the marginal percentage differences within

\mathcal{C}_{low}

and

\mathcal{C}_{high}

groups from one quarter to the next. Asterisks denote

p<0.05

Marker	Extreme				Neutral
Marker	P $\rightarrow$ C	Lag	C $\rightarrow$ P	Lag	P $\rightarrow$ C	Lag	C $\rightarrow$ P	Lag
Anger	3.20*	3	3.38*	3	1.32	4	0.81	4
Griev.	1.63	1	2.08	1	1.78	4	1.51	4
Power	2.77*	5	0.92	5	1.62	3	0.74	3
Neg.	0.49	3	0.25	3	0.96	5	2.58*	5
In-group	1.89	2	1.78	3	0.64	5	0.10	5

Table 3: The P

\rightarrow

C and C

\rightarrow

P columns show the F statistic for the SSR-based F-test in the Granger causality analysis, with the lag corresponding to the specification that yielded the lowest

p

-value among those reported. Asterisks denote

p<0.05

Panel of Figure 4 — Figure 4: Sensitivity analysis for data inclusion. Plot (a) illustrates the tradeoff between engagement filters and channel sample size, while (b) shows the impact of video count thresholds on participant retention (selected threshold $x=50$ marked in red).

Figure 5: Issue specific shifts in attitude scores over the year.

Opinion Dynamics

Opinion dynamics studies how people and agents influence one another’s beliefs, producing consensus, disagreement, or polarization.

See all 75 papers

When Words Divide: Diachronic Ideological Polarization in Political Discourse on Social Media
Aug 2, 2026Roy Yitzchak, Noa Lavie, Ella RabinovichSocial Media Analysis Political Discourse Analysis
The Academic College of Tel-Aviv Yaffo, Israel
Against Political Polarization: A Unified Framework for Tracing Evolving Political Ideologies on Social Media
Aug 18, 2026Yijie Xu, Chao Wang, Hui XiongSocial Media Analysis Political Discourse Analysis
Thrust of Artificial Intelligence, The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China · School of Artificial Intelligence and Data Science, University of Science and Technology of China, Hefei, China · Department of Computer Science and Engineering, The Hong Kong University of Science and Technology, Hong Kong, Hong Kong SAR, China
Scale over Preference: The Impact of AI-Generated Content on Online Content Ecology
Apr 2, 2026Tianhao Shi, Yang Zhang, Xiaoyan Zhao +8Social Media Analysis AI Governance
University of Science and Technology of China, Hefei, China. · National University of Singapore, Singapore. · Kuaishou Technology, Beijing, China. +1

Aug 2, 2026J/K move · Enter open · S save

When Words Divide: Diachronic Ideological Polarization in Political Discourse on Social Media

Roy Yitzchak, Noa Lavie, Ella Rabinovich

The Academic College of Tel-Aviv Yaffo, Israel

Social Media Analysis Political Discourse Analysis Computational Social Science

Open paper AskPDF arXiv

Political polarization has become a defining feature of online discourse, yet its long-term evolution remains poorly understood. We present a longitudinal analysis of ideological polarization in Reddit discussions by measuring semantic differences in the language used by opposing political communities. We construct temporally aligned community-specific word embeddings and quantify ideological polarization as the semantic divergence of political concepts over time. Our analysis shows that ideological polarization has increased substantially during the study period, both at the concept- and topic-level. Unlike prior computational work, which has largely focused on cross-sectional analyses or affective dimensions of polarization at a single point at time, our approach captures the evolution of ideological differences in semantic framing. The proposed framework provides a scalable method for studying the temporal dynamics of ideological polarization in large-scale social media discourse.

Abstract

Opinion Dynamics

Explore similar work

When Words Divide: Diachronic Ideological Polarization in Political Discourse on Social Media

Against Political Polarization: A Unified Framework for Tracing Evolving Political Ideologies on Social Media

Scale over Preference: The Impact of AI-Generated Content on Online Content Ecology