Introduction
The news media are often regarded as the fourth estate in democratic societies because of their role in monitoring the actions of government officials and exposing malfeasance (Besley and Prat Reference Besley and Prat2006; Snyder and Strömberg Reference Snyder and Strömberg2010; Larreguy, Marshall and Snyder Reference Larreguy, Marshall and Snyder2020). However, existing studies have cast doubt on such a view, arguing that the media could be influenced by the government and, therefore, may not be able to effectively promote accountability (Gehlbach and Sonin Reference Gehlbach and Sonin2014; Ozerturk Reference Ozerturk2022; Petrova and Enikolopov Reference Petrova, Enikolopov, Anderson, Waldfogel and Stromberg2015). Even in established democracies like the USA, where the freedom of the press is constitutionally guaranteed and the competition in the news market is intense (Gentzkow and Shapiro Reference Gentzkow and Shapiro2008; Besley and Prat Reference Besley and Prat2006), Qian and Yanagizawa-Drott (Reference Qian and Yanagizawa-Drott2017) find that ‘government distortion (of news coverage from independently owned outlets) has systematically existed’ (p. 467) – although, meanwhile, studying local daily newspapers in the USA from 1869 to 1928, Gentzkow et al. (Reference Gentzkow, Petek, Shapiro and Sinkinson2015) fail to find any evidence that the partisanship of the incumbent government impacts the partisan stance of the press.
Does the US government exert any influence on the US media? In this article, I extend this line of research by examining whether the US government shapes US newspaper coverage of foreign leaders and, if so, how. I focus on the coverage of foreign leaders rather than foreign countries, as scholars have found that Western media increasingly emphasize national leadership over the country as a whole in international reporting (Balmas and Sheafer Reference Balmas and Sheafer2013). Specifically, I examine the extent and the channel through which the US government influences domestic news outlets’ reporting on foreign leaders when the leaders violate human rights. Building upon existing scholarship in American politics and international relations, I argue that the US government would discourage media attention on human rights violators who are politically aligned with the country in order to avoid domestic criticisms and the resulting damages to US geopolitical interests. It would, however, encourage the news coverage of non-aligned human rights violators so as to ‘name and shame’ them and highlight the country’s commitment to human rights. I identify an important yet understudied channel of government influence: selective information provision during press briefings and through press releases, and hypothesize that, as a result, the relationship between foreign leaders’ human rights violations and US newspaper coverage of them would be contingent on the leaders’ political alignment with the USA. When leaders violate human rights, only those not aligned with the country would be covered more frequently.
To empirically examine this hypothesis, I collect yearly data on the frequency of mentions of foreign leaders across five major US newspapers: The New York Times, The Washington Post, The Wall Street Journal, The Chicago Tribune, and The Los Angeles Times from 1960 to 2015. Coupled with measures of human rights violations (Fariss Reference Fariss2019) and political alignment (Bailey, Strezhnev and Voeten Reference Bailey, Strezhnev and Voeten2017), this study offers the most comprehensive exploration to date of the factors shaping US newspaper coverage of foreign leaders and the bias in such coverage. I find strong and robust support for my hypothesis across various empirical specifications. That is, the effect of human rights violations on US newspaper coverage increases as the voting disparity with the USA in the United Nations General Assembly widens. Moreover, the conditional effect of human rights violations is more pronounced after the end of the Cold War, but is of similar magnitude across both liberal and conservative newspapers, irrespective of the incumbent administration’s partisanship.
While the above empirical pattern could be attributed to various factors, such as newspaper editors’ and journalists’ political preferences, I provide multiple pieces of evidence that the proposed channel of government influence – selective information provision during press briefings and through press releases – is an important factor in explaining the results. First, I demonstrate that the newspaper least reliant on the government for information abroad in my sample (by virtue of having the most foreign correspondents and foreign desks), namely The New York Times, shows the least bias in its coverage and no bias in its content about foreign leaders when the leaders commit human rights violations. Second, I utilize exogenous variation in countries’ geographical distances from the USA and conduct sub-sample analyses. I find that the coverage is biased only for distant countries, for which the government is more likely to be the media’s primary source of information due to geographical constraints. Finally, using transcripts of press briefings and press releases from the White House and the State Department, I provide direct evidence that the bias in US newspaper coverage of foreign leaders is at least partly driven by the government’s selective information provision.
This study draws on and contributes to scholarship on media bias and media capture within democratic contexts more broadly, and in the USA more specifically. Previous research on the US media has heavily emphasized the importance of demand-side factors (readers’ tastes) and supply-side factors (preferences of media owners, editors, and journalists) in driving the partisan bias in news outlets’ national and local reporting (for example, Hassell, Holbein and Miles Reference Hassell, Holbein and Miles2020; Puglisi and Snyder Reference Puglisi and Snyder2011; Gentzkow, Shapiro and Sinkinson Reference Gentzkow, Shapiro and Sinkinson2014). It has largely overlooked the non-partisan bias in news outlets’ international reporting.Footnote 1 This study fills this important gap by identifying not only the existence of such bias but also a key source of the bias – selective information provision by the government. Second, from an empirical standpoint, this research is among the first to leverage newspaper data to systematically explore the factors determining how frequently political leaders are featured in news reports.Footnote 2 This aspect is particularly important, given that such coverage, which has become increasingly prevalent, can easily shape public opinion of these leaders and their respective countries, affecting foreign relations and issues that transcend politics, including international trade (Balmas Reference Balmas2018). Finally, this study highlights that even independently owned outlets headquartered in established democracies can exhibit bias in favour of their home countries. It echoes Parkinson’s (Reference Parkinson2024) qualitative finding that ‘journalists do not operate to generate scientifically representative reporting’ (p. 4) and thus serves as an additional cautionary note against using event data sets constructed primarily or exclusively based on news articles.
What Drives US News Coverage of Foreign Leaders
The demand- and supply-side factors
Existing research posits that media coverage in democracies is governed by demand- and supply-side factors. On the demand side, news outlets are economically incentivized to respond to the preferences of consumers, who perceive like-minded information sources to be of higher quality and thus have greater demand for them (Gentzkow and Shapiro Reference Gentzkow and Shapiro2006; Reference Gentzkow and Shapiro2010, Gentzkow, Shapiro and Sinkinson Reference Gentzkow, Shapiro and Sinkinson2014). On the supply side, scholars have found that media owners, editors, and journalists, by virtue of their involvement in news production, can impose their preferences in deciding what and how stories are presented to readers (Larcinese, Puglisi and Snyder Reference Larcinese, Puglisi and Snyder2011; Martin and McCrain Reference Martin and McCrain2019; Hassell, Miles and Reuning Reference Hassell, Miles and Reuning2022).
These theories suggest that foreign leaders are likely to be covered more frequently in the news when they violate human rights. First, readers are likely to find news about human rights abuses and their perpetrators more engaging due to the inherent negative bias in human behaviour. That is, on average, individuals tend to pay more attention to and react more intensely to negative information compared to positive information (Baumeister et al. Reference Baumeister, Bratslavsky, Finkenauer and Vohs2001; Fournier, Soroka and Nir Reference Fournier, Soroka and Nir2020; Soroka, Fournier and Nir Reference Soroka, Fournier and Nir2019). For example, Trussler and Soroka (Reference Trussler and Soroka2014) find in a lab study supplemented by a survey that, readers, especially those with political interests, prefer negative news content when given a choice, regardless of their stated preferences. In response to this demand for negative news, media outlets are inclined to deliver more of it to maximize audience attention and revenue. Second, stories of human rights violators are also likely to captivate reporters because such stories are considered newsworthy by traditional journalistic standards. As a result, the media will mention foreign leaders more frequently when they violate human rights. Therefore, I first hypothesize the following:
Hypothesis 1: Without the US government’s influence, foreign leaders’ human rights violations increase US newspaper coverage of them.
The government’s influence
In the meantime, an increasing number of studies have pointed out that even in democratic countries, governments can capture the media through direct ownership, provision of financial resources, licensing and regulation, among other methods (Di Tella and Franceschelli Reference Di Tella and Franceschelli2011; Stanig Reference Stanig2015; Boas and Hidalgo Reference Boas and Hidalgo2011; Durante and Knight Reference Durante and Knight2012). Theoretically, there are two strategies through which the US government can influence domestic news outlets.
The first strategy is to apply direct pressure on news outlets by rewarding news editors and journalists who are cooperative with the administration’s agenda while punishing uncooperative ones (Qian and Yanagizawa-Drott Reference Qian and Yanagizawa-Drott2017). Such rewards may include exclusive interview opportunities with the President or senior officials, invitations to special briefings, or all-expenses-paid travel with the President (Grossman and Kumar Reference Grossman and Kumar1979). For punishments, the US government could restrict uncooperative news outlets’ future access to information (Ozerturk Reference Ozerturk2022),Footnote 3 launch personal attacks on uncooperative journalists, and pressure editorial boards to remove them from the field (Qian and Yanagizawa-Drott Reference Qian and Yanagizawa-Drott2017).Footnote 4
Given the country’s strong constitutional protections for press freedom, the second and more plausible strategy is to indirectly influence news outlets by selectively providing information to them. This can occur during press briefings and through press releases from the White House and the State Department, which reporters routinely attend and reference. Spokespersons and press secretaries could mention an entity – whether a country, a leader, or another figure – more frequently when the government’s goal is to increase media coverage of that entity and refrain from referring to it when the goal is to decrease coverage. This strategy would be particularly effective in shaping news outlets’ international reporting, as many news outlets have limited on-the-ground presence in foreign countries and often rely on government cues to determine where to allocate their scarce resources. Research on the media’s role in conflict warning and preventive policy has shown that, due to their limited foreign presence, the media often follow the government’s lead and only pay attention to local developments if the government has done the same (Otto and Meyer Reference Otto and Meyer2012). Furthermore, with the continuous decline in newsroom employment and, consequently, the media’s capacity to independently gather information (Peterson Reference Peterson2021), the government becomes increasingly capable of shaping news coverage by supplying media outlets with filtered information.
I hypothesize that, under the government’s influence, foreign leaders’ human rights violations will increase their news coverage frequency only if the leaders are not politically aligned with the USA. I argue that the US government has strong incentives to discourage media coverage of politically aligned leaders who violate human rights and encourage coverage of human rights violators who are not aligned. If news outlets extensively report on aligned leaders when they violate human rights, it would draw attention to the USA’s role in supporting, enabling, or, at the very least, tacitly condoning such behaviours. This can expose the US government, which has publicly committed to protecting and promoting human rights globally (US Mission Geneva 2016), to accusations of hypocrisy from opposition politicians and human rights activists. While such exposure may not critically harm a president’s re-election prospects, it could undermine their credibility, impact their ability to advance other policies, and tarnish their legacy (McManus and Yarhi-Milo Reference McManus and Yarhi-Milo2017). As demonstrated in McManus and Yarhi-Milo (Reference McManus and Yarhi-Milo2017), such concern about hypocrisy costs partly explains why the US government is more likely to offer discreet signals of support, such as military aid and arms sales, rather than more visible forms of support, like presidential visits to their autocratic allies. Moreover, the media’s naming and shaming of foreign leaders for their human rights abuses could sabotage important bilateral relationships and jeopardize US efforts in maintaining its strategic partners’ political alignment with the country, compromising the government’s ability to achieve its foreign policy objectives and geopolitical interests (Terman and Byun Reference Terman and Byun2022).
Conversely, extensive media reporting on human rights violations by leaders not aligned with the USA could damage these geopolitical rivals’ domestic and international legitimacy, potentially benefiting the USA (Lebovic and Voeten Reference Lebovic and Voeten2009). Such coverage can also help the US government build public support for opposing these leaders on the international stage or even taking action against them. As illustrated by Tomz and Weeks (Reference Tomz and Weeks2020), Americans view countries with poor human rights records as more threatening and are thus more inclined to support military actions against them. In addition, the President can leverage media reports highlighting non-alignment with human rights violators to underscore the country’s commitment to human rights, and relatedly, the consistency between their words and actions, a trait highly valued by the domestic audience (Levy et al. Reference Levy, Mckoy, Poast and Wallace2015). The argument resonates with Terman and Byun’s (Reference Terman and Byun2022) proposition that states face a thorny dilemma between ‘their image as global human rights promoters and their interests in important political relationships’ (p. 1). While they suggest states solve this dilemma by enforcing human rights norms selectively based on the target state’s sensitivity to these norms, this study focuses on how this dilemma influences the government’s interaction with the media.
There is also anecdotal evidence suggesting that the US media selectively covers leaders of foreign countries who violate human rights, consistent with the above argument. A telling example is the disparity in coverage frequency between leaders of Saudi Arabia and Venezuela (Supplementary Material Figure A1). Despite significant human rights issues, Saudi Arabia often receives less frequent and intense scrutiny in the US media due to its strong economic and strategic ties with the USA, including oil interests and regional security alignments. As noted by researchers from Human Rights Watch, there is ‘deafening US silence on Saudi rights’ (Coogle Reference Coogle2014). By contrast, Venezuelan leaders such as Hugo Chavez and Nicolas Maduro, who also preside over substantial human rights violations, are covered much more frequently and intensely due to the country’s adversarial stance towards the USA. Therefore, I also hypothesize the following:
Hypothesis 2: Under the US government’s influence, foreign leaders’ human rights violations increase US newspaper coverage of them only when the leaders are not politically aligned with the USA.
Empirical Design
Sample
To test my hypotheses, I first compile a sample of leaders (excluding US Presidents) who held power from 1960 to 2015. Following Goemans, Gleditsch and Chiozza (Reference Goemans, Gleditsch and Chiozza2009), leaders are defined as individuals who de facto exercise power (that is, effective primary rulers) in independent states. The initial list of leaders is sourced from the Archigos database (Goemans, Gleditsch and Chiozza Reference Goemans, Gleditsch and Chiozza2009), which I then cross-validate with data extracted from the REIGN (Rulers, Elections, and Irregular Governance) database (Bell, Besaw and Frank Reference Bell, Besaw and Frank2021). This results in a leader-year panel of 1,603 leaders from 179 countries, with leader-years exceeding ninety days in duration, that is, 9,284 observations.
Outcome variable: Newspaper coverage
To measure US newspaper coverage of foreign leaders, I search through the ProQuest Historical and Recent Newspapers database and programmatically collect the number of times five major US newspapers mention each leader in each leader-year.Footnote 5 The newspapers include The New York Times (NYT), The Washington Post (WP), The Wall Street Journal (WSJ), The Chicago Tribune (CT), and The Los Angeles Times (LAT).Footnote 6 Figure 1 shows the intensity of US media coverage of foreign leaders in 1960 and 2015.
US newspaper coverage of foreign leaders, 1960 and 2015.
Explanatory variables: Human rights violations and political alignment
I measure leaders’ human rights violations using the inverse of the country’s human rights practice score from Fariss (Reference Fariss2019), following the best-established approach (for example, Frantz et al. Reference Frantz, Kendall-Taylor, Wright and Xu2020; Hill and Jones Reference Hill and Jones2014; Jones and Lupu Reference Jones and Lupu2018). The score is derived from an aggregation of event-based and standards-based data on state-sanctioned repression, including political imprisonment, torture, and extrajudicial killings, from multiple sources. These data are combined using a latent variable model and thus account for the systematic changes in how reporting agencies like Amnesty International encounter and interpret information about human rights abuses (Fariss Reference Fariss2014).
Leaders’ political alignment with the USA is approximated by their country’s ideal point distance from the USA in the United Nations General Assembly (UNGA). This measure is constructed by calculating the absolute difference in ideal points between the country and the USA, with the ‘ideal points’ representing national foreign policy preferences estimated using a spatial theory IRT model based on countries’ voting in the General Assembly (Bailey, Strezhnev and Voeten Reference Bailey, Strezhnev and Voeten2017). Scholars have used the same or similar measures (correlation in the UNGA votes, the fraction of the UNGA votes voted with the USA, etc.) for the same purpose of measuring political alignment or, so to speak, ‘geopolitical affinity’ (for example, Alesina and Dollar Reference Alesina and Dollar2000; Qian and Yanagizawa-Drott Reference Qian and Yanagizawa-Drott2009; Terman and Byun Reference Terman and Byun2022). In the data, higher numbers indicate lower levels of political alignment between the leader and the USA.
Including both human rights violations and political alignment measures in the same model naturally raises concerns about multicollinearity. One might argue that human rights violators are less likely to be politically aligned with the USA, while leaders who respect human rights are more likely to be aligned. However, these two variables exhibit only moderate correlation, with a coefficient of 0.419. There are several historical examples of leaders, such as Carlos Arana Osorio in Guatemala, António de Oliveira Salazar in Portugal, and Benjamin Netanyahu in Israel, who were known human rights violators but were also considered to be close US allies. To further address this concern, alternative model specifications are provided in the Supplementary Material. In Table B1, the variables are dichotomized to minimize their correlation. In Table B2, the effect of human rights violations on US newspaper coverage is estimated separately for politically aligned and non-aligned leaders.
Model specification
To test my first hypothesis, I estimate an ordinary least-squares (OLS) regression model that takes the following form:
$${Y_{i,c,t+1}} = {\beta _1}Human\,Rights\,\,Viol{._{c,t}} + {\beta _2}UNGA\,Ideal\,Pt.Dist{._{c,t}} + {\bf{X}}^{'}\theta + {\mu _{i,c}} + {\nu _t} + {\varepsilon _{i,c,t}}$$
where
$i$
,
$c$
,
$t$
index leader, country, and year, respectively. The outcome variable,
${Y_{i,c,t+1}}$
, is the average number of times leader
$i$
of country
$c$
is mentioned by newspapers in year
$t+1$
(in log).Footnote
7
The variables
$Human\,Rights\,Viol{._{c,t}}$
and
$UNGAI\,deal\,Pt.Dist{._{c,t}}$
represent the measures of human rights violations and political alignment, respectively. To test the second hypothesis, which posits that the effect of human rights violations on US news coverage is conditional on leaders’ political alignment with the USA, I add the interaction of the two variables (
$Human\,Rights\,Viol{._{c,t}} \times UNGAI\,deal\,Pt.Dist{._{c,t}}$
) to the model. If the hypothesis holds, the coefficient of this interaction term would be positive and statistically significant. This would suggest that US newspaper coverage of foreign leaders is influenced not only by the leaders’ human rights violations but also by their degree of political alignment with the USA.
The vector of control variables,
${\bf{X}}$
, includes several variables that might confound the relationship between US media coverage, human rights violations, and political alignment. The first set of variables captures country
$c$
’s domestic conditions, such as GDP per capita, GDP growth, population size, oil wealth (measured by oil production per capita), levels of democracy (polyarchy), presidentialism,Footnote
8
media freedom, and political instability.Footnote
9
Binary variables indicating whether an election occurs for the leader’s office and whether the country experiences intra- or inter-state conflict are also included. The second set of control variables captures the country’s strategic value to or relationship with the USA. It includes a binary variable indicating whether the country holds a seat on the United Nations Security Council on a rotating basis, whether the country is a US military ally, measures of the country’s trade with and aid from the USA, and binary variables indicating whether the leader receives a visit from the US President or is invited to visit the USA. Summary statistics and data descriptions are provided in Supplementary Material Tables A1 and A2.
Although I do not seek to make a causal claim in this article, I adopt several strategies to improve identification. First, the outcome variable is led forward to address potential concerns regarding reverse causality. Second, I use leader fixed effects,
${\mu _{i,c}}$
, to account for unobserved heterogeneity in average newspaper coverage across leaders, and year fixed effects,
${\mu _t}$
, to account for time-specific changes in US newspaper coverage of foreign entities.Footnote
10
Finally, the standard errors are clustered at the leader level to adjust for serial correlation in newspaper coverage within leaders.
Results
Main results
Table 1 presents the main results on the determinants of US newspaper coverage of foreign leaders. I first run a parsimonious model that includes only the explanatory variables (Columns 1 and 4), then sequentially incorporate the first (Columns 2 and 5) and the second (Columns 3 and 6) sets of control variables, as previously discussed. Leader and year fixed effects are incorporated in all models. When not interacting with the UNGA ideal point distance, human rights violations appear to increase US newspaper coverage in a sizable and statistically significant way. A one-standard-deviation increase in human rights violations is associated with an 18.8 per cent (
$0.148 \times 1.269$
) to 24.1 per cent (
$0.190 \times 1.269$
) increase in the number of times leaders are mentioned by the US newspaper in the following year. Nevertheless, once the interaction term is added, the coefficient for human rights violations is no longer statistically distinguishable from 0. The positive and significant interaction term indicates that the effect of human rights violations on US newspaper coverage increases as the UNGA ideal point distance to the USA grows (that is, political alignment with the USA decreases).
Human Rights Violations, Political Alignment, and US Newspaper Coverage of Foreign Leaders
Notes: Robust standard errors clustered at the leader level are reported in parentheses. + p<0.10, *p<0.05, ** p<0.01, ***p<0.001.
The conditional marginal effect of human rights violations on US newspaper coverage is graphically depicted in Figure 2. I use a kernel smoothing estimator that allows for more flexible estimations of the functional form (Hainmueller, Mummolo and Xu Reference Hainmueller, Mummolo and Xu2019). The figure once again confirms that US newspapers primarily react to human rights violations of leaders who are not politically aligned with the USA. For aligned leaders, their human rights violations do not lead to a statistically significant increase in coverage. Nonetheless, this also suggests that aligned human rights violators are covered less frequently than they otherwise would be – since in normal cases, leaders’ human rights violations should increase their coverage, regardless of their political alignment with the USA. To address the concern about the non-linear interaction effect, in Supplementary Material Table B1, I dichotomize the two explanatory variables and re-run the above model; in Supplementary Material Table B2, I estimate the effect of human rights violations on US newspaper coverage for politically aligned and non-aligned leaders separately. The results remain substantively unchanged.
Marginal effect of human rights violations.
Notes: Leader and year fixed effects are included. Robust standard errors clustered at the leader level.
Several additional findings from Table 1 are worth noticing. First, American newspapers appear to have a stronger inclination toward covering negative news from foreign countries. There exists a strong, positive, and statistically significant relationship between political instability and US newspaper coverage. However, analyses in Supplementary Material Table C1 and Figure C1 reveal that this relationship is also moderated by leaders’ political alignment with the USA. This is in line with my theory, which by extension, would predict that the US government would not want to make the public aware that leaders sharing its foreign policy preferences are unpopular domestically.
Second, there is mixed evidence on the relationship between US geopolitical interests and newspaper coverage. On the one hand, leaders from oil-rich countries or countries that are military allies of the USA tend to receive more newspaper coverage. On the other hand, neither holding a temporary seat at the United Nations Security Council (a position shown to be of great value to the USA by Kuziemko and Werker Reference Kuziemko and Werker2006) nor having strong economic ties with the USA (measured by imports and exports) exerts any statistically significant influence on US newspaper coverage. Receiving more economic or military aid from the USA also does not have any impact. Finally, I find no evidence that a state visit to the USA or from the US President heightens newspaper attention in the upcoming year – this is hardly surprising, given that the effect of diplomatic exchange on newspaper coverage is unlikely to be persistent.Footnote 11
Robustness checks
I conducted a number of robustness tests and auxiliary analyses, with their findings detailed in the Supplementary Material. I account for missing data with multiple imputations rather than list-wise deletion (Table B3) (Lall Reference Lall2016),Footnote
12
use the outcome variable in year
$t$
instead of
$t+1$
(Table B4), run Poisson regressions to mitigate the problem of a zero-inflated outcome variable (Table B5) (Wooldridge Reference Wooldridge1999), estimate two fully moderated models to address potential omitted interaction bias (Table B6 and Table B7) (Blackwell and Olson Reference Blackwell and Olson2022), exclude fixed effects from the model (Table B8), and estimate models with mixed effects (Table B9). The conditional effect remains sizable and statistically significant in all cases. In most cases, the effect size is similar to that in the original analysis. Moreover, Tables B10 to B13 show that the results are not driven by any specific subgroup of leaders, such as non-US allies who frequently violate human rights (for example, China, Russia, and Iran). My second hypothesis is strongly supported.
Alternative explanations
Next, I consider a range of alternative explanations for the above results.
First, there may be omitted variables. To test for this, I conduct sensitivity analyses to assess how severe the confounding would have to be to substantially alter the results (Cinelli and Hazlett Reference Cinelli and Hazlett2020). I find that even a confounder ten times stronger than oil production per capita or intrastate conflict would not eliminate the interaction effect (Figure D1). Moreover, in Table 2, I replace the outcome variable with the average number of times a leader and ‘human rights’ are jointly mentioned by newspapers (in the log). The results provide additional support for my hypothesis. It becomes even clearer that foreign leaders’ human rights violations per se have little impact on their coverage in US newspapers; their effect is fully moderated by the leaders’ political alignment with the USA. The conditional marginal effect of human rights violations on US newspaper coverage of foreign leaders’ human rights practices is graphically depicted in Supplementary Material Figure D2.Footnote 13
Human rights violations, political slignment, and US newspaper coverage of foreign lader’s human rights practices
Notes: Robust standard errors clustered at the leader level are reported in parentheses.
$^ +p \lt 0.10$
, *
$p \lt 0.05$
, **
$p \lt 0.01$
, ***
$p \lt 0.001$
.
Second, it could be argued that the finding is driven by the effect of US newspaper coverage on the human rights practices of foreign leaders, conditional on the leaders’ political alignment. However, if this argument were valid, the estimates would be biased downward (that is, the interaction effect would be underestimated), since by heightening the possibility of international sanctions (especially for non-aligned leaders), US newspaper coverage is more likely to improve, rather than worsen foreign leaders’ respect for human rights (Peksen, Peterson and Drury Reference Peksen, Peterson and Cooper Drury2014). To address this concern empirically, I have led the outcome variable for one year to reduce the likelihood of reverse causality (Table 1). Additionally, in Supplementary Material D2, I show that when two lags of the outcome variable are incorporated into the model, the size and statistical significance of the interaction term remain similar.
Next, if the measure I employ overrates (underrates) aligned (non-aligned) leaders’ human rights violations, it would also appear that the US newspapers under-report (over-report) aligned (non-aligned) leaders when they violate human rights. This, however, is unlikely. If anything, the measure is likely biased in the opposite direction (that is, underrating aligned leaders’ human rights violations and overrating non-aligned leaders’ human rights violations), given that some of its underlying indices, including the CIRI physical integrity data, Hathaway torture data, Political Terror Scale (PTS), PITF data on genocide and politicide, are constructed partly or entirely based on information from the Country Reports on Human Rights Practices published by the US Department of State. Therefore, any bias in the measure of human rights violations would cause an under-estimation rather than an over-estimation of the effect size.
The fourth consideration pertains to whether the findings could be attributed to the differences in the types of human rights violations committed by different leaders. It might be argued that leaders who are not politically aligned with the USA tend to commit more stigmatized ‘newsworthy’ forms of human rights abuses, such as torture and killing, whereas aligned leaders may violate human rights by failing to protect citizens’ socio-economic rights or the rights of women and children, which attract far less media attention. It is important to note that the human rights violation measure I use exclusively captures violations of physical integrity rights (Fariss Reference Fariss2019). There is no reason to expect significant variations in the newsworthiness of political imprisonment, torture, extrajudicial killings, etc., as they all involve ‘grievous bodily harm while directly implicating state authorities as willful abusers’ (Terman and Byun Reference Terman and Byun2022).
Finally, could biased US newspaper coverage of foreign leaders result from demand-side factors, specifically news outlets catering to the tastes of their audience? This would be true if American readers were more interested in enemies’ human rights violations than those of allies. To examine this possibility, I collect Google Trends (GT) data for searches on foreign leaders in the USA, which I then use as a proxy for US public attention to these leaders. The underlying assumption is that the average American’s interest in a foreign entity can be inferred from the frequency of searches for that entity on Google, the country’s most widely used search engine. The GT data, accessed through Google’s Application Programming Interface (API), captures the popularity of a term (that is, a foreign leader) in relation to itself over the time frame in a given location (that is, the USA) (Timoneda and Wibbels Reference Timoneda and Wibbels2022). Therefore, I normalize the GT score of a foreign leader within his tenure. Leader fixed effects are kept in the model to address the concern that GT scores are not comparable across leaders. Supplementary Material Table D3 first confirms that there is a positive and statistically significant correlation between US newspaper coverage of foreign leaders and public attention to these leaders. However, Supplementary Material Table D4 finds that the bias in favour of politically aligned human rights violators and against non-aligned ones exists only in the former and is not present in the latter, suggesting that readers’ demand is unlikely to be the driver of the bias in newspapers.
Effect heterogeneity
To explore the heterogeneity in the conditional effect of human rights violations on US newspaper coverage of foreign leaders, I first divide the data into two periods: the Cold War and post-Cold War eras (Columns 1–4, Supplementary Material Table E1). Then, I interact human rights violations and UNGA ideal point distance with the partisanship of the incumbent US President (Columns 5–8, Supplementary Material Table E1). Additionally, I disaggregate the outcome variable to examine whether liberal (conservative) newspapers are more biased under Democratic (Republican) administrations (Supplementary Material Table E2).Footnote 14 The results reveal that the government’s influence over the media is more pronounced in the post-Cold War era. However, I find no statistically significant evidence suggesting a systematic difference in the coverage of foreign leaders between liberal and conservative newspapers, or between Democratic and Republican administrations.
Overall, the results are consistent with the idea that the US government’s influence on domestic news outlets transcends partisanship – as both Democratic and Republican administrations have vested interests in shaping how newspapers of different political leanings cover leaders who are politically aligned and unaligned with the USA. The results arguably also speak against the argument that these newspapers self-curate their news coverage to cater to the political interests of the incumbent administration. If that were the case, the conditional effect of human rights violations on US newspaper coverage would be more pronounced for liberal newspapers during Democratic administrations and for conservative newspapers during Republican administrations.
Before discussing the mechanisms, it is important to consider another aspect of effect heterogeneity: whether the conditional effects of human rights violations on US newspaper coverage of foreign leaders vary by the type of coverage being examined.Footnote 15 Importantly, news coverage carries tone. When foreign leaders violate human rights, could the shifts in the amount of positive and negative coverage differ depending on whether the leaders are politically aligned with the US or not? To answer this question, I use The New York Times’ Application Programming Interface (API) to programmatically access and download over 250,000 articles on foreign leaders from the newspaper and conduct a dictionary-based sentiment analysis. Following standard practice, I use the Lexicoder Sentiment Dictionary (LSD) (Young and Soroka Reference Young and Soroka2012) and measure the tone of each title by the percentage point difference between positive and negative words (Soroka, Stecula and Wlezien Reference Soroka, Stecula and Wlezien2015). Articles with sentiment scores above the average are classified as positive, while those below the average are categorized as negative. I then count the number of each type for each leader-year. The results are presented in Supplementary Material Tables E3 and E4, with the key findings summarized in Table 3.
Human rights violations, political alignment, and The New York Times’ coverage of foreign leaders
Notes: Robust standard errors clustered at the leader level are reported in parentheses.
$^ + $
$p \lt0.10$
, *
$p \lt 0.05$
, **
$p \lt 0.01$
, ***
$p \lt 0.001$
.
The analyses reveal that an increase in human rights violations is associated with a decrease in positive coverage (Columns 1–3, Table E3), with this effect being uniform for both politically aligned and non-aligned leaders (Columns 1–3, Table 3). However, the effect of human rights violations on the amount of negative coverage foreign leaders receive depends on their political alignment with the USA; the less aligned the leaders are, the stronger the effect of their violations (Columns 4–6, Table 3). In other words, when non-aligned leaders violate human rights, they experience a similar decrease in positive coverage as aligned leaders but face a disproportionately larger increase in negative coverage. This indicates that the bias in (The New York Times) coverage of foreign leaders is primarily a bias in negative coverage. However, it is important to note that these results are based on data from a single newspaper. It is still theoretically possible that other newspapers may show bias in both their positive and negative coverage of foreign leaders. For example, the negative effect of human rights violations on the amount of positive coverage could be less pronounced for politically aligned leaders compared to non-aligned leaders in these newspapers.Footnote 16 Moreover, as the next section will demonstrate, The New York Times does not exhibit bias against non-aligned leaders in terms of either total coverage or average tone. Therefore, readers should approach the generalizability of the findings above with caution.
Mechanism
So far, this paper has presented strong empirical evidence that US newspaper coverage of foreign leaders is biased in favour of politically aligned human rights violators and against non-aligned ones. What accounts for this bias? While there could be multiple explanations, this study argues that the US government’s selective information provision during press briefings and through press releases significantly contributes to this phenomenon. In the following section, I provide several pieces of evidence supporting this claim.
First, if the bias is (at least partly) driven by the government’s selective information provision, news outlets that are less dependent on the government for information abroad would be less biased. Among the five newspapers examined, it is reasonable to assume that The New York Times is the newspaper least dependent on the government’s information provision as it has the most extensive global outreach. It employs the largest number of foreign correspondents, operates the most overseas desks,Footnote 17 and has received the highest number of Pulitzer Prizes for International Reporting.Footnote 18 Supplementary Material Table F1 thus lends initial support to the mechanism of selective information provision. The estimates indicate that the effect of human rights violations on US newspaper coverage is moderated by the political alignment of foreign leaders with the USA in all newspapers except The New York Times. Admittedly, the preceding analyses show that The New York Times is biased in its negative coverage of foreign leaders, but it appears that the newspaper achieves a degree of balance overall. This is further supported by analyzing the newspaper’s average tone of coverage, calculated as the mean sentiment score of all articles related to a leader within a given leader-year. As demonstrated in Supplementary Material Table F2, increased human rights violations lead to a decline in the tone of coverage of foreign leaders, irrespective of the leaders’ political alignment.
This evidence is obviously indirect. The New York Times and other newspapers in my sample are systematically different in aspects beyond their global outreach, which could account for the above findings. Therefore, I introduce the second piece of evidence that utilizes exogenous variation in countries’ geographical distances from the USA. If the US government exerts influence on news outlets through selective information provision, the coverage of leaders from more distant countries – where news outlets face greater geographical constraints in establishing on-the-ground presence and are thus more reliant on the government for information – would be more biased. Supplementary Material Table F3 splits the data into two sub-samples using the country’s minimal distance to the USA (below and above the median) and estimates the conditional effect of human rights violations on US newspaper coverage for the two sub-samples separately. In line with the mechanism of selective information provision, only the coverage of leaders from distant countries is biased. For neighbouring countries, the effect of human rights violations on US newspaper coverage is not conditional on their leaders’ political alignment with the USA.
As the final and perhaps most important step in probing the proposed mechanism of government influence, selective information provision, I collect the transcripts of press briefings and press releases of the White House and State Department from the archived websites of the Bush and Obama administrations.Footnote 19 I count the number of times each leader was mentioned each year from 2001 to 2015. The results are presented in Table 4.Footnote 20 Columns 1–3 reaffirm the conditional effect of human rights violations on US newspaper coverage in this period. Columns 4–6 confirm that US government coverage of foreign leaders is similarly biased. The effect of leaders’ human rights violations on the White House and the State Department also increases as the leaders’ political alignment decreases (that is, UNGA ideal point distance increases). Finally, Columns 7–9 show a strong, positive, and statistically significant relationship between government coverage and newspaper coverage, and that the conditional effect of human rights violation on newspaper coverage loses its statistical significance once the amount of government coverage is controlled for.
Human rights violations, political alignment, and US newspaper and government coverage of foreign leaders
Notes: Robust standard errors clustered at the leader level are reported in parentheses.
$^ + $
$p \lt 0.10$
, *
$p \lt 0.05$
, **
$p \lt 0.01$
, ***
$p \lt 0.001$
.
It is important to emphasize that, in highlighting the importance of the government’s selective information provision in shaping US newspaper coverage of foreign leaders, I am not dismissing the possibility that the US government may have other ways to influence domestic news outlets. Nor am I suggesting that other factors, such as editors’ and journalists’ political preferences, do not matter in explaining the biased newspaper coverage of foreign leaders. It is entirely possible that editors and journalists themselves want to devote more coverage to human rights violators who are not politically aligned with the USA so as to defend and advance US national interests. While I cannot fully rule out this alternative explanation, the empirical evidence suggests that supply-side factors alone are insufficient to account for the results. Otherwise, we would likely observe similar bias in The New York Times’ coverage and in newspaper coverage of leaders from neighbouring countries – since there is no reason to expect that NYT editors and journalists are less interested in US national interests, or that editors and journalists would only impose their political preferences while covering distant countries and not the neighbouring ones. Moreover, the conditional effect of human rights violations on US newspaper coverage would retain its statistical significance even when controlling for the White House and the State Department’s coverage. Granted, none of the evidence presented above causally identifies the effect of selective information provision on newspaper coverage, and each piece of evidence has its respective flaws. However, taken together, they consistently point toward the same conclusion that the biased coverage of foreign leaders is at least partly driven by the US government’s selective information provision to news outlets, which, unfortunately, often rely on the government for information in international reporting.
Discussion and Conclusion
Governments worldwide seek to influence the stories reporters write. This article advances the research on media capture in democracies by analyzing whether the US government influences domestic news outlets’ coverage of foreign leaders, and, more importantly, how it does so. I hypothesize and empirically demonstrate that the extent to which human rights violations lead to increased newspaper coverage depends on foreign leaders’ political alignment with the USA. When leaders infringe upon human rights, those aligned with the USA in the United Nations General Assembly often escape the media spotlight, while those who vote contrary to US interests garner more attention than they otherwise would. I attribute this pattern (partly) to the US government’s selective information provision, a channel of government influence that has not been explored in existing literature. To provide empirical evidence, I first demonstrate that the newspaper with the most extensive global outreach, and thus least reliant on the US government for overseas information, namely The New York Times, shows the least bias in its coverage and no bias in its content about foreign leaders. Then, through sub-sample analyses, I demonstrate that the bias in newspaper coverage of foreign leaders primarily occurs in the coverage of leaders from geographically remote countries, where news outlets, due to geographic constraints, are more dependent on government-provided information. Lastly, using documents from the White House and State Department, I show directly that the US government affects newspaper coverage of foreign leaders through selective information provision during press briefings and in press releases.
The findings that US newspaper coverage of foreign leaders is biased and that this bias is at least partly driven by the US government’s selective information provision, carry significant implications. It has long been believed that the news media wield power over foreign policy making through their ability in agenda-setting, priming, and framing. Wood and Peake (Reference Wood and Peake1998), for instance, demonstrate that the media can change ‘public perceptions of the relative importance of foreign policy problems’, thereby disturbing the orientation of presidential attention. Soroka (Reference Soroka2003) documents the media’s effects on the public salience of foreign affairs, which prompt governmental responses. Nevertheless, this study highlights the inverse scenario, in which the government sets the news media’s agenda. This implies not only that the media, particularly in the realm of international reporting, are susceptible to the influence of political elites, but also that their ability to bolster government accountability in foreign policy making may be limited, as they cannot provide citizens with unbiased information to make well-informed decisions.
It is important to note several caveats. First, this study is not a critique of the US media or government. It would be unreasonable to expect any government to indiscriminately provide information about its allies and non-allies to the press, or for any media outlet to rely solely on its own resources for international reporting. That being said, this suggests that similar government influences on the media may be pervasive in countries besides the USA, where incumbents frequently hold press conferences to relay information to the public. Data from a single case, unfortunately, do not permit further speculation; future research is required to better understand the generalizability of these findings to non-US contexts. Second, this study does not identify the causal effects of the US government on domestic news outlets, nor does it find that the bias in US newspaper coverage of foreign leaders should be fully attributed to the US government. While it offers evidence that the government’s selective information provision is an important factor in explaining the biased coverage, it cannot completely rule out all alternative explanations. Other factors, including editors’ and journalists’ political preferences, could also contribute to the bias. Even the US government may have other ways to influence the media (Qian and Yanagizawa-Drott Reference Qian and Yanagizawa-Drott2017). More research is needed to fully understand what drives American newspapers’ bias in favour of politically aligned leaders and against non-aligned ones. Finally, this study focuses on the association between political alignment and bias in newspaper coverage. An intriguing avenue of research would be to examine the bias in broadcast media and study whether leaders’ strategic value or military importance to the USA also shapes the leaders’ coverage. These investigations could yield more insights into the complex dynamics between media reporting and international politics.
Supplementary material
Supplementary material for this article can be found at https://doi.org/10.1017/S0007123424000991.
Data availability statement
Replication data for this article can be found in the Harvard Dataverse at https://doi.org/10.7910/DVN/383IMF.
Acknowledgements
I thank Jean Hong, Deniz Aksoy, Quintin Beazer, Taylor Carlson, Ted Enamorado, Xincheng Ge, Li Han, Ning He, Siqin Kang, Franziska Barbara Keller, Victoria Liu, Christopher Lucas, Jacob Montgomery, William Nomikos, Diana O’Brien, Anna Wilke, Wenhui Yang, Han Zhang, participants at workshops at the Hong Kong University of Science and Technology and Washington University in St. Louis, and panel participants at the 2021 American Political Science Association Annual Meeting and 2022 Midwest Political Science Association Annual Meeting for their constructive feedback on earlier drafts of this paper. I also thank the editor, Ruth Dassonneville, and three anonymous reviewers for their helpful comments. All errors are my own.
Financial support
None.
Competing Interests
None.
Open access
