Media Slant is Contagious

Philine Widmer; Clémentine Abed Meraim; Sergio Galletta; Elliott Ash

econ.GNFeb 15, 2022

Media Slant is Contagious

Authors: Philine Widmer, Clémentine Abed Meraim, Sergio Galletta, Elliott Ash

Organizations: Paris School of Economics · ETH Zürich

Abstract

This paper examines the diffusion of media slant. We document the influence of Fox News Channel (FNC) on the partisan slant of local newspapers in the U.S. over the years 1995-2008. We measure the political slant of local newspapers by scaling the news article texts to Republicans' and Democrats' speeches in Congress. Using channel positioning as an instrument for viewership, we find that higher FNC viewership causes local newspapers to adopt more right-wing slant. The effect emerges gradually, only several years after FNC's introduction, mirroring the channel's growing influence on voting behaviour. A main driver of the shift in newspaper slant appears to be a change in local political preferences.

Figures & tables

Figure 1: Democrat (Light Grey) and Republican (Dark Grey) Bigrams, 2005-2008

Figure 2: First Stage: Fox News Viewership and Channel Position

Figure 3: Reduced Form: Republican Newspaper Slant and Fox News Position

	Newspaper Slant
	(1)	(2)	(3)
Fox News Channel Viewership	0.718**	0.493***	0.515**
	(0.276)	(0.169)	(0.204)
Observations	718	718	718
K-P First-Stage F-Stat	9.01	14.70	15.06
State Fixed Effects	✓	✓	✓

Table 1: 2SLS: Newspaper Slant and Fox News Viewership

	Republican
	Newspaper Slant
	Average of Score
	(1)	(2)	(3)
Fox News Channel Viewership	0.655*	0.310	0.414**
	(0.341)	(0.191)	(0.192)
Observations	718	718	718

Table 2: 2SLS: Newspaper Slant and Fox News Viewership Robustness: Slant as Average of Raw Scores

Figure 4: Reduced Form: (Historical) Republican Newspaper Slant and Fox News Position

Appendix figures & tables33 assets

Supplementary material from the paper’s appendix.

Appendix

All periods:
Filtering	# Articles	Results
Raw scrapes 1995-2008	93,731,393	None (impossible: no
		county assignment)
Main sample 2005-2008:
Raw scrapes	43,495,343	None (impossible: no
		county assignment)

Appendix

Table A.1: Number of Articles Collected and Filtering

Variable	Mean	STD	Min	Max	N
Newspapers
05-08 main slant measure (share Rep. articles)	0.36	0.07	0.00	1.00	718
05-08 alternative slant (as main but incl. obituaries)	0.35	0.07	0.00	1.00	718
05-08 alternative slant (avg. article-level partisanship)	0.00	0.03	-0.39	0.36	718
05-08 main slant only for local news	0.35	0.07	0.00	1.00	718
05-08 Democrat slant (share Dem. articles)	0.33	0.06	0.00	1.00	718

Appendix

Table A.2: Summary Statistics

Variable	Mean	STD	Min	Max	N
Demographics
Population (1000s)	411	1048	6	9819	718
Population density	675	2410	3	33886	718
Republican vote share 1996	0.43	0.10	0.14	0.68	718
Share Whites	0.79	0.16	0.21	0.99	718
Share Blacks	0.10	0.12	0.00	0.77	718

Appendix

Table A.2 Continued: Summary Statistics

	Predicted Democrat	Predicted Republican
Actual Democrat	31.58% (58K)	23.25% (43K)
Actual Republican	12.92% (24K)	32.26% (59K)

Appendix

Table B.1: Prediction Performance for Identifying Congressional Speeches - GST Model 2005-2008

Model Period	2005-2008	2001-2004	1998-2000	1995-1997
Number of Congress Speeches	182,943	171,852	143,747	171,637
Share of Republican Speeches (%)	45.2	46.7	55.1	55.4

Appendix

Table B.2: Number of congress speeches per model

Period	Democrat	Coef	Count	Republican	Coef	Count
2005–2008	gun safeti	-3.55	579	liabil reform	3.51	105
2005–2008	surplus trillion	-2.69	27	margin tax	3.05	125
2005–2008	gun violenc	-2.51	1806	unborn child	2.92	1709
2005–2008	break wealthi	-2.50	101	crime introduc	2.89	18
2005–2008	redeploy troop	-2.48	197	size govern	2.82	261
2005–2008	occup iraq	-2.45	1006	medic liabil	2.78	322

Appendix

Table B.3: Most partisan bigrams by period model

Figure B.1: Wordcloud of top partisan bigrams over 2005-2008 - Fontsize represents frequency

Figure B.2: Distribution of Slant Across Newspapers

Field	Filter Criteria
Title	Contains: “death notice”
	Contains: “obituaries”
	Contains: “obituary”
	Contains: “notice of sale under power”
	Exact match: “births”
Paragraph	Starts with: “births”

Appendix

Table B.6: Routine Announcement Filtering Criteria

Local	Coef	Non-local	Coef
local	10.71	dear	-4.45
counti	10.03	fy	-3.93
polic	5.87	american	-3.77
area	5.84	bush	-3.63
communiti	5.42	china	-3.59
gov	5.18	ny	-3.58

Appendix

Table B.7: Most Predictive Unigrams for Local Classification

	Fox News Viewership
	(1)	(2)	(3)	(4)	(5)	(6)
Fox News Channel Position	-0.098	-0.142***	-0.173***	-0.104	-0.152***	-0.185***
	(0.068)	(0.046)	(0.044)	(0.081)	(0.051)	(0.048)
MSNBC Channel Position			0.043			0.040
			(0.026)			(0.027)
Observations	718	718	718	718	718	718

Appendix

Table C.1: First Stage: Cable TV Position Effects on Viewership

	Predictions for:
		Republican
	Fox News Viewership	Newspaper Slant
	(1)	(2)
Fox News Channel Position	0.006	0.001
	(0.054)	(0.025)
Observations	718	718

Appendix

Table C.2: Identification Checks: Fox News Position Correlation with Covariates

	Republican
	Newspaper Slant
	All Outlets
	(1)	(2)	(3)	(4)
Fox News Channel Viewership	0.128*	0.097	0.072	0.089
	(0.074)	(0.079)	(0.083)	(0.068)
Households with FNC Access [Share]			2.993	2.233

Appendix

Table C.3: OLS: Fox News Viewership Effects on Newspaper Slant

	Republican
	Newspaper Slant
	(1)	(2)	(3)
Fox News Channel Position	-0.111***	-0.091***	-0.094**
	(0.027)	(0.031)	(0.047)
MSNBC Channel Position		0.057**	0.018
		(0.028)	(0.024)

Appendix

Table C.4: Reduced Form: Newspaper Slant and Fox News Position

	Republican
	Newspaper Slant
	(1)	(2)	(3)
MSNBC Viewership	-1.153	-1.399*	-0.420
	(0.847)	(0.765)	(0.578)
Observations	718	718	718
K-P First-Stage F-Stat	10.84	13.01	14.13

Appendix

Table C.5: 2SLS: Newspaper Slant and Fox News Viewership

	Republican
	Newspaper Slant
	Partisan Articles Only
	(1)	(2)	(3)
Fox News Channel Viewership	0.702**	0.424**	0.559***
	(0.317)	(0.204)	(0.207)
Observations	717	717	717

Appendix

Table C.6: 2SLS: Newspaper Slant and Fox News Viewership Robustness: Slant Only Relative to Partisan Articles

	Democrat
	Newspaper Slant
	(1)	(2)	(3)
Fox News Channel Viewership	-0.364	-0.019	-0.199
	(0.347)	(0.185)	(0.185)
Observations	718	718	718
K-P First-Stage F-Stat	9.01	14.70	15.06

Appendix

Table C.7: 2SLS: Newspaper Slant and Fox News Viewership Robustness: Democrat Slant

	Republican
	Newspaper Slant
	Outlets $\geq$ 1000 Art.
	(1)	(2)	(3)
Fox News Channel Viewership	0.456*	0.550**	0.481*
	(0.249)	(0.236)	(0.262)
Observations	541	541	541

Appendix

Table C.8: 2SLS: Newspaper Slant and Fox News Viewership Robustness: Outlets with

\geq

1000 Articles

	Republican
	Newspaper Slant
	All Articles
	(1)	(2)	(3)
Fox News Channel Viewership	0.596**	0.450***	0.470**
	(0.261)	(0.164)	(0.196)
Observations	718	718	718

Appendix

Table C.9: 2SLS: Newspaper Slant and Fox News Viewership Robustness: Keeping Routine Announcements (e.g., Obituaries)

	Republican
	Newspaper Slant
	(1)	(2)	(3)
Fox News Channel Viewership	0.831*	0.461*	0.449*
	(0.445)	(0.243)	(0.229)
Observations	608	608	608
K-P First-Stage F-Stat	5.67	10.11	10.35

Appendix

Table C.10: 2SLS: Newspaper Slant and Fox News Viewership Robustness: Historical Circulation

	Republican
	Newspaper Slant
	(1)	(2)	(3)
Fox News Channel Viewership	0.707**	0.591**	0.625***
	(0.322)	(0.222)	(0.229)
Observations	718	718	718
K-P First-Stage F-Stat	8.90	17.50	17.61

Appendix

Table C.11: 2SLS: Newspaper Slant and Fox News Viewership Robustness: Relative Circulation

	Republican
	Newspaper Slant
	(1)	(2)	(3)
FNC Viewership relative to MSNBC	1.085***	0.875***	0.916**
	(0.384)	(0.305)	(0.411)
Observations	718	718	718
K-P First-Stage F-Stat	16.96	10.72	10.66

Appendix

Table C.12: 2SLS: Newspaper Slant and Fox News Viewership Robustness: Fox News Relative to MSNBC Viewership

	Republican
	Newspaper Slant
	(1)	(2)	(3)
Fox News Channel Viewership	0.776**	0.656***	0.657***
	(0.315)	(0.221)	(0.216)
Observations	558	558	558
K-P First-Stage F-Stat	7.38	13.96	14.50

Appendix

Table C.13: 2SLS: Newspaper Slant and Fox News Viewership Robustness: Dropping Counties with Low Nielsen Respondent Number

	Republican
	Newspaper Slant
	(1)	(2)	(3)
Fox News Channel Viewership	0.782***	0.659***	0.467**
	(0.267)	(0.196)	(0.178)
Observations	718	718	718
K-P First-Stage F-Stat	9.51	15.61	17.07

Appendix

Table C.14: 2SLS: Newspaper Slant and Fox News Viewership Robustness: Census Division Fixed Effects

	Newspaper Slant
	(1)	(2)	(3)
Fox News Channel Viewership	0.718*	0.493*	0.515**
	(0.368)	(0.267)	(0.257)
Observations	718	718	718
State Fixed Effects	✓	✓	✓
Demographic Controls	✓	✓	✓

Appendix

Table C.15: 2SLS: Newspaper Slant and Fox News Viewership Robustness: Robust Standard Errors

Panel of Figure 32 — Figure C.1: 2SLS: Newspaper Slant and Fox News Viewership Robustness: Dropping States Individually

Figure D.1: Reduced Form: (Historical) Republican Newspaper Slant and Fox News Position Robustness: Time-Varying Controls

Figure D.2: Reduced Form: (Historical) Democrat Newspaper Slant and Fox News Position

Figure D.3: Reduced Form: Predicted Treatment Effects on Slant and Republican Votes in 2008

Strong Response of Slant	Strong Response of Voting
Share age 30-59	Share males
Share Hispanics	Share age 30-59
Share service occupations	Share arts occupations
Share office occupations	Median tax rate

Appendix

Table D.1: Covariates Most Associated with High Treatment Effects (County-Level)

		Republican
	Local	Newspaper Slant
	Share	Local Articles
	(1)	(2)	(3)	(4)
Fox News Channel Viewership	-0.045	0.628**	0.381**	0.423**
	(0.038)	(0.270)	(0.150)	(0.185)
Observations	718	718	718	718

Appendix

Table D.2: 2SLS: Newspaper Slant and Fox News Viewership Mechanisms: Slant in Local News Articles

	Newspaper
	Circulation
	(1)	(2)	(3)
Fox News Channel Position	-0.011	0.013	0.009
	(0.063)	(0.061)	(0.062)
MSNBC Channel Position		0.002	0.008
		(0.034)	(0.030)

Appendix

Table D.3: Reduced Form: Circulation and Fox News Position Mechanisms: Effects on Circulation

	Newspaper
	Circulation
	(1)	(2)	(3)
Fox News Channel Position	0.011	0.013	0.012
	(0.030)	(0.033)	(0.033)
MSNBC Channel Position		-0.014	-0.014
		(0.035)	(0.036)

Appendix

Table D.4: Reduced Form: Circulation and Fox News Position Mechanisms: Effects on Circulation

Political Communication

See all 27 papers

A Multi-View Media Profiling Suite: Resources, Evaluation, and Analysis
May 2, 2026Muhammad Arslan Manzoor, Dilshod Azizov, Daniil Orel +4Multi-View Feature Fusion Multi-View Learning
αMBZUAI, UAE · γUniversity of Texas at San Antonio, USA · δUniversity of Copenhagen, Denmark +1
Does Welsh media need a review? Detecting bias in Nation.Cymru's political reporting
Apr 19, 2026Cai Parry-JonesPolitical Communication
University of Cambridge, Cambridge, United Kingdom
NewsLens: A Multi-Agent Framework for Adversarial News Bias Navigation
May 17, 2026Joy BoseMulti-Agent Systems
Independent Researcher, Bengaluru, India

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

A Multi-View Media Profiling Suite: Resources, Evaluation, and Analysis

Muhammad Arslan Manzoor, Dilshod Azizov, Daniil Orel, Umer Siddique+3

αMBZUAI, UAE · γUniversity of Texas at San Antonio, USA · δUniversity of Copenhagen, Denmark+1

Multi-View Feature Fusion Multi-View Learning Text Classification

Open paper AskPDF arXiv

News outlets shape public opinion at a scale that makes automated detection of political bias and factuality essential. However, the field still lacks unified resources, comprehensive evaluations across diverse approaches, and systematic analyses of the representations and fusion strategies that matter most, especially under label sparsity and dataset diversity. In addition, there is little empirical work reporting broad, observation-driven findings about what consistently works, what fails, and why. We address these gaps through four main contributions. First, we introduce MBFC-2025, a large-scale label set covering approximately 2,600 outlets from Media Bias/Fact Check (MBFC). Second, we construct multiview representations for ACL-2020 (Panayotov et al., 2022), which includes around 900 outlets, as well as for MBFC-2025. These representations span Alexa graphs, hyperlink graphs, LLM-derived graphs, articles, and Wikipedia descriptions. Third, we provide a systematic evaluation and analysis of embedding views and fusion strategies, including a reinforcement learning-based fusion variant. Fourth, we conduct extensive experiments that achieve state-of-the-art results on ACL-2020 and establish strong benchmarks on MBFC-2025.

Abstract

Political Communication

Explore similar work

A Multi-View Media Profiling Suite: Resources, Evaluation, and Analysis

Does Welsh media need a review? Detecting bias in Nation.Cymru's political reporting

NewsLens: A Multi-Agent Framework for Adversarial News Bias Navigation