Research

Informational autocrats rely on the media to generate support, yet they face a critical obstacle: regime opponents tend to select out of exposure to government media. This paper argues that autocrats circumvent selective exposure by controlling media outlets through proxy owners with concealed ties to the state. Drawing on original ownership investigations and surveys in four African countries, I show that proxy outlets are now more prevalent than traditional state broadcasters and that most citizens are unaware that these outlets are owned by regime actors. In preregistered field experiments in Tanzania, I demonstrate how such knowledge gaps benefit autocrats. When ownership is disclosed, regime opponents avoid proxy outlets and seek alternative sources. When ownership is concealed, however, opponents consume proxy outlets as readily as regime supporters do, and are persuaded by their pro-regime coverage. The findings challenge the notion that propaganda can only affirm existing regime supporters, highlighting how autocrats use concealment to reach and persuade even skeptical audiences.

There is growing awareness of how autocrats use propaganda to generate and sustain mass support, yet much of this attention focuses on the influence of directly state-owned media. This article shifts the spotlight to proxy-owned media—private outlets controlled by individuals with informal ties to autocratic regimes. In autocracies around the world, proxy media are on the rise, overtaking traditional state-run outlets as the most prevalent and far-reaching source of propaganda. Unlike state media, proxy outlets maintain a veneer of political independence, making them appear more trustworthy to skeptical audiences. This article highlights gaps in public knowledge about the true ownership of proxy outlets, explores how these gaps may make otherwise skeptical people more likely to consume and believe proregime news, and suggests ways journalists, activists, and the public can counter proxy media's influence.

Whether propaganda influences mass attitudes has implications for the ability of authoritarian regimes to maintain power, yet studies have reached varying conclusions about its effectiveness. We present a meta-reanalysis of 108 authoritarian propaganda experiments across 37 countries, covering messages justifying war and repression, praising autocrats' performance, and vilifying their rivals. Three key findings emerge. First, propaganda is effective on average: it shifts policy views, improves performance evaluations, increases regime support, and reduces protest willingness. Second, effects are strikingly consistent across subgroups, including regime supporters and non-supporters, the educated and uneducated, and the politically engaged and disengaged, challenging prominent theories about individual differences in susceptibility to propaganda. An exception is that foreign audiences respond more than domestic ones, though the distribution of stimuli differs systematically across audiences. Third, we find limited evidence of heterogeneity by message content. Conditional on exposure, propaganda is generally effective, underscoring the ample power autocrats have to shape public opinion.

Bardia Rahmani [first author], Jiayi Zhang, Alexander Coppock, and Jennifer Pan

We describe a natural experiment occasioned by an abrupt increase in the transmission range of an independent Tanzanian radio station whose broadcasts emphasize current affairs and gender equality. Some villages that formerly lay outside the catchment area of this radio station could now receive it, while nearby villages remained outside of reception range. Prior to the change in transmitter range in 2018, we conducted a baseline survey in both treated and untreated villages and found them to be similar in terms of prevailing social attitudes and political interest. An endline survey conducted in 2020 shows that respondents in areas that received the new radio signal were substantially more likely to listen to the station, and their levels of political interest and knowledge about domestic politics were significantly higher than their counterparts in villages where the signal could not reach. Attitude change on a range of gender issues, however, was sporadic.

Donald P. Green, Dylan W. Groves, Constantine Manda, Beatrice Montano, and Bardia Rahmani

Media outlets in competitive autocracies are often semi-independent: formally free of state control, yet prone to self-censorship and reliant on government sources. Although such outlets are common, research has focused on fully independent or government-owned media. This paper explores how semi-independent media shape public opinion among urban, opposition-leaning audiences. In a randomized field experiment in Tanzania, participants received one month of free access to 31 national newspapers via a mobile app. Treated participants substantially increased their consumption of semi-independent newspapers. In turn, they became more politically knowledgeable, but also developed more favorable attitudes toward the regime on average. An analysis of heterogeneity suggests that effects depended on citizens' prior media consumption: semi-independent media increased pro-regime attitudes among citizens who ordinarily consume critical online news, and increased knowledge among those who consume less-critical traditional outlets. Semi-independent media may thus serve autocrats' interests by persuading those citizens most skeptical of the regime.

Early and forced marriage (EFM) is an increasing focus of international organizations and local non-government organizations. This study assesses the extent to which attitudes and norms related to EFM can be changed by locally tailored media campaigns. A two-hour radio drama set in rural Tanzania was presented to Tanzanian villagers as part of a placebo controlled experiment randomized at the village level. A random sample of 1,200 villagers was interviewed at baseline and invited to a presentation of the radio drama, 83% of whom attended. 95% of baseline respondents were reinterviewed two weeks later, and 97% fifteen months after that. The radio drama produced sizable and statistically significant effects on attitudes and perceived norms concerning forced marriage, which was the focus of the radio drama, as well as more general attitudes about gender equality. Fifteen months later, treatment effects diminished, but we continue to see evidence of EFM-related attitude change. 

Donald P. Green, Dylan W. Groves, Constantine Manda, Beatrice Montano, and Bardia Rahmani

Is narrative entertainment simply a form of recreation, or does it have meaningful effects on public opinion? Building on prior reviews, we present a meta-analysis of 377 findings from 77 experiments evaluating the persuasive effects of narrative radio, television, and film, including a growing body of work from low- and middle-income countries. Our sample includes both entertainment-first narratives — popular media created primarily to entertain but which may incidentally shape audiences’ attitudes, beliefs, and behaviors — and education-first narratives designed by policymakers to inform, persuade, or motivate public action. Using a hierarchical-effects model, we assess narrative media’s influence across a wide range of settings and issue domains. The results suggest that narrative entertainment is quite influential, with sizable persuasive effects that remain apparent weeks after initial exposure. A smaller literature reports head-to-head tests of the relative effectiveness of narrative versus non-narrative messages; although inconclusive, the evidence suggests that narratives may be only slightly more persuasive than non-narrative messages. If true, this finding would imply that the main advantage of narratives may be their ability to attract and engage large and diverse audiences. We conclude by calling attention to gaps in the literature and proposing avenues for further research.

Bardia Rahmani [first author], Donald P. Green, Dylan W. Groves, and Beatrice Montano

Do rulings by high courts influence social attitudes and perceived norms? Evidence from outside of industrialized democracies is scarce. This paper investigates the influence of a recent Tanzanian high court decision that laws permitting marriage for girls under the age of 18 are unconstitutional. We begin by showing that just 8% of rural Tanzanians had heard about the ruling one year after it was made. We then report the results of a lab-in-the-field experiment to test whether media reports about the court's decision influence attitudes, perceived norms, and intentions to report early marriage to authorities. We randomly assigned 1,950 respondents in rural Tanzania to hear a radio news story about the ruling or to a control condition. Respondents who heard the report were 8 percentage points more likely to reject early marriage and 5 percentage points more likely to say they would report early marriage to authorities, but no more likely to believe their rural community rejects early marriage or to volunteer to speak out publicly against early marriage. The effect of the report attenuated but did not disappear when it included the information that Tanzania's Attorney General disagreed with the court decision. We conclude that high court rulings can influence public opinion, but these effects depend on dissemination and framing by news media. 

Salma Emmanuel, Dylan W. Groves, Constantine Manda, Donald P. Green, Beatrice Montano, Bardia Rahmani

Foreign investors frequently collude with domestic political elites to acquire communal land in the Global South, with negative consequences for local livelihoods and ecosystems. These "land grabs" often succeed because elites provide communities with one-sided information that overstates benefits and downplays risks. This paper tests a scalable counter-messaging intervention: a one-hour radio drama that depicts villagers learning about—and mobilizing against—a corrupt deal between a foreign developer and a local official to sell off communal lands. We evaluate the intervention using a placebo-controlled cluster-randomized field experiment in rural Tanzania in which 1,360 respondents from 34 villages were randomly invited to attend a screening of the radio program. One month later, respondents in treated villages were significantly more likely to prioritize environmental protection over large-scale development projects, reject a politician calling to sell off local lands, and vote for a conservationist candidate in mock elections. The findings show that media campaigns can improve accountability by challenging elite justifications for rent-seeking behavior. 

Bardia Rahmani [first author], Dylan W. Groves, Beatrice Montano, and Donald P. Green

Can religious leaders use their standing to promote women's political participation (WPP) in the developing world? We present the results of experiments conducted in northeastern Tanzania that estimate the effect of a pro-WPP audio message from a progressive religious elite on behavioral intentions, attitudes, and norms relating to female office-seeking. Across two studies (N = 4,701), we find that the progressive religious elite's message makes villagers more likely on average to say they would encourage their daughter or niece to run for political office. These effects on behavioral intentions persisted a month later. Moreover, we find suggestive evidence that the children of respondents who received the progressive elite's message became more interested in running for political office. By contrast, we find muted or inconsistent effects of the progressive message on attitudes and norms, and of conservative messages on all three outcomes. The results suggest that interventions that leverage local religious elites can effectively reduce familial gatekeeping of women's political participation in socially conservative communities.

Bardia Rahmani [first author], Dylan W. Groves, Beatrice Montano, and Francis Ngatigwa

What is the effect of climate-focused media on public opinion in developing countries? We conducted a survey of climate change knowledge, attitudes, and behavioral intentions among 496 respondents in rural northeastern Tanzania. We find while 70% of respondents recognize that the environment is declining, less than 57% of respondents are aware of the negative impact of climate change and just 16% are aware that climate change is primarily caused by activities outside of Tanzania. To address this knowledge gap, we developed two 2-minute radio news reports in collaboration with a local Tanzanian radio station Pangani FM that describe the causes and consequences of climate change. In the first clip, the radio presenter emphasizes local responsibility for environmental degradation; in the second clip, the newscaster specifically attributes responsibility for climate change to China and the United States. We randomly assign respondents to listen to one of the two news vignettes or to a pure control condition. On average, the treatments dramatically increased respondents' understanding of the human causes and negative consequences of climate change and moderately increased their willingness to enforce natural resource governance rules, but did not influence attitudes and perceived norms about balancing environmental conservation and economic self-interest. We also find suggestive evidence that the news clip emphasizing international causes of climate change decreased willingness to engage in local efforts to protect the environment relative to the domestic causes condition.

Can edutainment sensitize listeners to the problem of gender-based violence (GBV) and build support for a collective response? While a robust literature focuses on the incidence of intimate partner violence in Sub-Saharan Africa, few studies consider the broader range of threats that women experience in public settings, such as harassment, sexual assault, and sexual violence. We study how edutainment shapes awareness, policy priorities, and preferred responses to these aspects of GBV through a placebo-controlled experiment randomized at the village level in rural Tanzania. A random sample of 1,250 villagers was interviewed at baseline and invited to one of two randomly assigned radio drama screenings, then interviewed again one month later.  The 90-minute radio drama that focuses on GBV both raises awareness about the risks women face in their daily lives and increases the importance that audiences accord to sexual violence as a community problem. Narrative mass media offers an effective and scalable means for spurring collective action responses to threats to women's safety in public spaces.

Beatrice Montano, Donald P. Green, Dylan W. Groves, and Bardia Rahmani

People weigh electoral quality more heavily than any other factor when considering whether a country is democratic or not. However, we know relatively little about how people evaluate whether an election meets democratic standards. Drawing on a conjoint experiment implemented in ten countries, we study which kinds of electoral violations are most likely to be perceived as serious and the contexts in which they are most likely to trigger opposition. Results indicate that violations affecting voting are perceived as most damaging to electoral quality, whereas violations that restrict candidate competition or distort information are viewed as less problematic. In addition, we show evidence that violations do not consistently provoke stronger reactions when targeting the opposition specifically, and that respondents update about the seriousness of violations based on electoral outcomes. The results suggest that people in different contexts hold relatively similar views about what determines electoral quality, and they highlight vulnerabilities that may allow authoritarian political leaders to use elections to bolster claims of democratic legitimacy. 

with Eddy Yeung  (Oxford), Lisa Fan (Oxford), Jonathan Chu (National University of Singapore), and Scott Williamson (Oxford)

Development assistance often requires cooperation with governments that repress their populations, creating tensions for practitioners charged with advancing climate protection, poverty alleviation, or governance reform. These tensions become especially acute when governments employ repression in service of implementing development projects themselves. How do development professionals weigh project benefits against state-inflicted harms, and do they discount those harms when projects align with their organizational mandates? To examine these questions, we field an elite experiment with practitioners working in major development agencies, international NGOs, and government ministries across three sectors: climate, poverty reduction, and governance. We present respondents with a series of profiles of real-world projects in which governments have repressed individuals residing in areas targeted for large-scale development. The conjoint design randomizes (1) the project’s stated purpose (climate action, poverty alleviation, or a non-development control), (2) information about the government’s conduct (a government statement emphasizing consultation with affected people; a human rights report documenting forced evictions, arrests, and violence; or a no-information control), and (3) whether affected communities are described as an “ethnic group” or as “indigenous people.” Outcome measures capture whether respondents support the project, how much funding they would allocate from a hypothetical budget, and how they evaluate the implementing government’s democratic credentials and suitability as a development partner. This design will allow us to identify whether practitioners exhibit sectoral bias -- tolerating repression when projects advance their own organizational interests -- and how frames communicated by governments and human rights organizations shape such decisions. 

with Janina Beiser-McGrath (Royal Holloway) and Liam Beiser-McGrath (LSE)

High-profile revelations that China and Russia have paid journalists in the Global South to publish favorable news stories have raised concerns about the vulnerability of national media to foreign influence. Yet the hidden nature of media bribery makes it notoriously difficult to study. Are most journalists willing to place political content on behalf of external actors or is the practice confined to a small minority? Does payment increase journalists’ receptivity? And are journalists in certain contexts or types of outlets more susceptible than others? We investigate these questions through a global audit experiment. Using an original dataset of media practitioners across multiple regions, we contact journalists with requests to publish news articles, randomly varying both the content (neutral versus pro-China, pro-Russia, or pro-U.S. articles) and the offer (no payment versus escalating payments). We record responses but do not proceed with any placements. This design allows us to estimate the causal effect of payments on willingness to run political propaganda, describe the frequency with which journalists acquiesce to such requests, and identify the institutional and contextual conditions under which journalists are most and least susceptible to external influence. 

with Erin Baggott Carter (USC, Stanford) and Brett Carter (USC, Stanford)

For nearly a century, scholars have studied the ways in which exposure to mass communication influences social and political attitudes. Although experimental studies abound, they primarily focus on how exposure to a particular program or channel affects what audiences think; rarely have experiments randomly provided the means to consume a particular type of mass media. The present the results of two experiments in rural Tanzania in which survey respondents were randomly assigned to receive radios and were re-interviewed approximately 15-20 months later.  Although the treatment group, as expected, became much more likely to listen to radio than the control group, the downstream effects on political and social attitudes are mixed. We find no appreciable effects on political participation or interest in politics, but we do see gains in knowledge about current events and increased concern about crime.  Attitudes about gender equality tend to move in the more progressive direction, and those who received radios also tend to become more accepting of stigmatized groups.  Those in the treatment group came to rate the ruling party more favorably, perhaps reflecting the political capture of nominally independent radio stations in the region, but these effects are not large and do not apply to ratings of the current president.

with Donald P. Green, Dylan W. Groves, and Beatrice Montano