Who Rallies Round the Flag? The Impact of the US Sanctions on Iranians’ Attitude toward the Government

Abstract

While politicians often argue that economic sanctions can induce policy changes in targeted states by undermining elite and public support for the reigning government, the efficacy of these measures, particularly against non-democratic regimes, is debatable. We propose that, counterintuitively, economic sanctions can bolster rather than diminish support for the sanctioned government, even in non-democratic contexts. However, this support shift and its magnitude can differ across various political factions and depend on the nature of the sanctions. To empirically evaluate our theoretical expectations, we use supervised machine learning to scrutinize nearly 2 million tweets from over 1,000 Iranian influencers, assessing their responses to both comprehensive and targeted sanctions during Donald Trump’s presidency. Our analysis shows that comprehensive sanctions generally improved sentiments toward the Iranian government, even among its moderate oppositions, rendering them more aligned with the state’s stance. Conversely, while targeted sanctions elicited a milder rally-around-the-flag response, the identity of the targeted entity plays a crucial role in determining the scale of this reaction.

Bien que les politiques affirment souvent que les sanctions économiques peuvent impulser des changements politiques dans les États ciblés en ébranlant le soutien des élites et du public au gouvernement en place, l'efficacité de ces mesures, notamment contre des régimes non démocratiques, est discutable. Nous proposons que, contre toute logique, les sanctions économiques puissent renforcer, et non diminuer, le soutien en faveur du gouvernement sanctionné et ce, même dans les contextes non démocratiques. Néanmoins, cette évolution du soutien et son ampleur sont susceptibles de varier d'une faction politique à l'autre et dépendent de la nature des sanctions. Pour évaluer empiriquement nos attentes théoriques, nous avons recours à l'apprentissage automatique supervisé afin d'analyser près de 2 millions de tweets de plus de 1 000 influenceurs iraniens. Nous évaluons ainsi leur réaction aux sanctions générales et ciblées durant le mandat de Donald Trump. D'après notre analyse, les sanctions générales rehaussent souvent l'estime portée au gouvernement iranien, même chez son opposition modérée, et donc l'alignent davantage sur la position de l’État. À l'inverse, tandis que les sanctions ciblées suscitent une légère réaction de ralliement au drapeau, l'identité de l'entité prise pour cible joue un rôle crucial lorsqu'il s'agit de déterminer l'ampleur de cette réaction.

Si bien los políticos argumentan, con frecuencia, que las sanciones económicas pueden provocar cambios a nivel de política en algunos Estados específicos debido a que debilitan el apoyo público y de las élites al Gobierno en el poder, la eficacia de estas medidas, particularmente contra regímenes no democráticos, es discutible. Sugerimos que, en contra de lo que se podría intuir, las sanciones económicas pueden ayudar reforzar el apoyo al Gobierno sancionado, incluso en contextos no democráticos, en lugar de disminuirlo. Sin embargo, este cambio en materia de apoyos, así como su magnitud, pueden diferir entre las distintas facciones políticas y depender de la naturaleza de las sanciones. Con el fin de estudiar de manera empírica nuestras expectativas teóricas, utilizamos el aprendizaje automático supervisado para examinar casi 2 millones de tuits escritos por más de 1000 «influencers» iraníes y analizamos sus respuestas a las sanciones, tanto integrales como selectivas, que tuvieron lugar durante la presidencia de Donald Trump. Nuestro análisis demuestra que las sanciones integrales, en general, mejoraron los sentimientos hacia el Gobierno iraní, incluso entre sus opositores moderados, provocando una mayor alineación de estos con la postura del Estado. Por el contrario, si bien las sanciones selectivas provocaron una respuesta más moderada con respecto a este apoyo, la identidad de la entidad objetivo desempeña un papel crucial con respecto a la determinación de la escala de esta reacción.

On May 8, 2018, US President Donald Trump reintroduced economic sanctions against Iran by withdrawing from the 2015 Joint Comprehensive Plan of Action (JCPOA), more widely recognized as the Iran Nuclear Deal (Landler 2018). The reinstated sanctions ignited discussions among American politicians and pundits about their potential efficacy in inducing policy change within the Iranian government. Concurrently, the potency of economic sanctions has long been contested in academia. Critics highlight two primary factors undermining the success of sanctions: the prevalence of state repression and the “rally round the flag” effect (Wood 2008; Peksen 2009; Drezner 2011).

The micro-foundational understanding of the “rally round the flag” effect has traditionally been limited due to the dearth of individual-level data, especially within non-democratic contexts where data accessibility is a salient concern. To address this limitation, our study develops a theoretical typology explaining how various groups of citizens respond to different economic sanctions. We argue that the emergence of a “rally around the flag” effect depends on citizens’ attitudes toward the status quo and their preferred approach to maintaining or changing it. As an implication of our proposed typology, foreign policy events like economic sanctions can exacerbate divisions among opposition groups, deepening their internal cleavages and reducing their unity. This fragmentation undermines the effectiveness of both comprehensive and targeted sanctions intended to exert pressure on the state by mobilizing opposition. Instead, it can lead to disagreement over the legitimacy and fairness of the sanctions, ultimately hindering their ability to achieve the intended political outcomes. This study emphasizes the importance of considering the strategic reactions of citizens within opposition groups in response to different types of sanctions.

This study also contributes to an emerging line of scholarship that capitalizes on online platforms, specifically Twitter,¹ to elucidate conflict dynamics and citizen behavior in authoritarian settings (King, Pan, and Roberts 2013; Han 2015; Gohdes 2018, 2020; Jost et al. 2018; Homburg 2019; Larson et al. 2019). Leveraging natural language processing and machine learning techniques, we examine a collection of tweets from Iranian influencers to draw nuanced inferences about individual reactions to state policies and international pressures. By analyzing this effect in the specific context of sanctions, our study not only improves our understanding of the mechanisms driving the rally phenomenon but also emphasizes the growing relevance and utility of social media data in advancing social science research. This approach exemplifies the innovative means by which digital data, when paired with computational methods, can complement the current literature to yield richer, more comprehensive insights into political behaviors and attitudes.

In the context of rising international tensions, nations’ approach to addressing perceived threats remains a critical area of study. Specifically, during Donald Trump’s presidency, the US–Iran relationship saw heightened tensions, mainly stemming from concerns over the Islamic Republic’s nuclear program and its regional ambitions. In addressing these, the Trump administration initiated a maximum pressure strategy, manifested in a series of comprehensive and targeted sanctions against Iran. This paper capitalizes on this unique circumstance, where varied sanctions were imposed within a relatively short period, to delve into and compare the nuanced impacts of these policies on a nation. Such an exploration is critical in understanding the broader implications of sanction-driven foreign policies, especially in today’s geopolitical landscape.

This article is structured as follows: We first review the existing literature on economic sanctions, emphasizing their role as a non-violent tool in foreign policy. Then, we develop our theoretical framework, connecting different types of economic sanctions to the rally around the flag effect across different groups of citizens. Subsequently, we discuss our research design and methodologies. We then present our empirical findings and engage in a thorough discussion. Finally, we conclude by summarizing our findings, addressing the study’s limitations, and suggesting avenues for future research.

Sanctions as a Foreign Policy Tool

Historically, economic sanctions have been perceived as a non-violent and humane strategy to resolve international disputes. Powerful governments often employ sanctions, or the threat thereof, to incentivize policy shifts in adversarial nations. By pressurizing a country’s elites and the general populace, sanctions increase the socio-political and economic costs associated with a state’s status quo, prompting reconsideration of policies (Bapat et al. 2013, 84). In non-democratic settings, the economic strain from sanctions can spark protests against the state (Allen 2008a, 2008b). To counteract this, state authorities typically resort to coercive measures. While such repression escalates human rights abuses (Wood 2008; Peksen 2009; Peksen and Drury 2010; Liou, Murdie, and Peksen 2021), it often effectively stifles dissent, resisting externally driven policy directives (Drezner 2011). Protests triggered by sanctions and subsequently met with state repression can structurally increase the state’s repressiveness (Davenport and Appel 2022), and the heightened fear of severe repression increases the costs of any future mobilization efforts.

Several scholars also have found that economic sanctions can increase the likelihood of terrorism in targeted states. On the one hand, targeted governments may blame foreign states for the economic hardships caused by sanctions, potentially leading to a higher risk of terrorist attacks against foreign citizens (Choi and Luo 2013). On the other hand, economic sanctions, coupled with state repression, can radicalize opposition groups, elevating the risk of terrorist attacks from these groups (Heffington 2017; McLean et al. 2018). The escalation of conflict and increase in human rights violations are often considered “unintended” consequences of economic sanctions (Eriksson 2010; Meissner and Mello 2022), yet these outcomes have been extensively studied by scholars in the field and should not be considered unanticipated (Zwart 2015).

The scholarship on economic sanctions has also suggested that “the rally around the flag” effect can be another reason for the failure of these foreign policies to achieve their goals (Galtung 1967; Cortright and Lopez 2000; Sejersen 2021). Mueller (1970, 20) coined the term “rally round the flag” while studying the periodic rise in support for an incumbent US president during major international events involving the United States. Since then, this effect, where foreign crises boost incumbent governments’ support, has been extensively studied in democratic contexts (e.g., Stimson 1976; Oneal and Bryan 1995; Kobayashi and Katagiri 2018). Yet, scholars have also found evidence of rallying around the flag during international crises in authoritarian states. For instance, Hale (2018) found that Russia’s annexation of Crimea boosted Putin’s popularity by strengthening patriotic sentiment. In another study, Hellmeier (2021, 450) showed that international pressure, such as economic sanctions, tends to increase pro-government mobilization in support of authoritarian leaders. These studies indicate that foreign crises can strengthen a sense of national unity, leading to greater backing for incumbent regimes. This rallying in support of the incumbent government can play a significant role in shaping the public’s response to international sanctions, suggesting that such measures might not always have the intended destabilizing effect on targeted states, even in authoritarian states.

The literature indicates that both repressed protests and the “rally around the flag” effect are observed outcomes and may contribute to the failure of economic sanctions. Wood (2008) has argued that, indeed, sanctioned populations often oscillate between dissent and rallying. This implies that assuming the entire population reacts to sanctions in a monolithic way—either opposing the state due to economic hardships or rallying behind it out of patriotic sentiments—is overly simplistic. To explain macro-level trends and observations, we need a more nuanced, micro-level understanding of these dynamics. Despite its significance, the “rally around the flag” effect has received relatively little attention in the academic discourse on economic sanctions, where most discussions focus on the dissent-repression cycles. Studying this oscillation in authoritarian regimes is indeed complex. Authoritarian states may stage pro-state demonstrations to exaggerate their support during challenging times, creating an illusion of broad public approval. At the same time, fear of repression (Ritter and Conrad 2016) can deter opposition groups from protesting, leading to an absence of visible dissent, which can be misinterpreted as support for the regime. Thus, it is challenging to discern whether the reduced opposition against the state results from an increased fear of repression due to sanctions or from the opposition’s reluctance to appear unpatriotic by maintaining pressure on the incumbent government during an international crisis.

To avoid these pitfalls, it is crucial to analyze individual-level data to understand how different citizen groups perceive economic sanctions and what drives their strategic responses. The motivations of these groups can vary depending on their attitude toward the status quo and the type of sanctions imposed. By incorporating a micro-level perspective, we can capture the variations in reactions and identify the underlying mechanisms that can drive the rally around the flag effect across different groups in non-democratic states. Our research indeed aligns with recent scholarship that emphasizes the importance of understanding the mechanisms behind the success or failure of sanctions (Early and Cilizoglu 2020). Through developing a theoretical framework and micro-level analysis of individuals’ online behavior, this study seeks to explain who rallies around the flag in non-democratic states by explaining how different types of sanctions lead to varying strategic responses among both pro- and anti-government citizen groups.

Theoretical Discussions

The literature identifies two prevalent sanction types (Peksen 2019; Özdamar and Shahin 2021): comprehensive and targeted. While comprehensive sanctions broadly exert pressure on the targeted state to induce policy changes, their political repercussions and limited success in non-democratic contexts have given rise to more nuanced, targeted sanctions (Drezner 2011, 96). Targeted sanctions, often called “smart sanctions” (Cortright and Lopez 2000), pinpoint specific politicians and their supporting elites and organizations, thereby minimizing collateral civilian impact. Theoretically, as targeted sanctions avoid broadly affecting citizens, it is perceived that they are less prone to instigate widespread protests, subsequent state repression, or invoke the “rally round the flag” sentiment. Instead, the aim is for the sanctions to target the benefits and interests of the winning coalition (Smith, Siverson, and Morrow 2003) in the targeted state, thereby prompting them to exert pressure on political leaders and elites to alter their policies.

However, even within the winning coalition of authoritarian regimes, politicians and elites do not constitute a monolithic group (Weeks 2009, 2012). These coalitions often include a variety of factions with differing levels of commitment to the regime and attitude toward the status quo. The rally around the flag effect, therefore, can hinge on the specific target chosen by the sanctioning entity. If the targeted elite or organization is perceived as a potential internal agent of change within the incumbent regime, then moderates are likely to disapprove of sanctions against them. Conversely, if the targeted sanctions focus on elites and factions vital to the state’s repressive machinery, the opposition is more likely to support such measures, leading to a reduced likelihood of a rally around the flag effect.

For example, if sanctions are aimed at a moderate or reformist elite of an authoritarian regime who might advocate for changes, moderate oppositions are likely to view such sanctions unfavorably. In this context, sanctions may appear as an impediment to internal reform, leading moderate oppositions to see them as unfair and mistargeted, leading to a higher likelihood of rallying around the government in opposition to foreign interference. On the other hand, if the sanctions target the hardliner elements within the regime—such as security forces, military leaders, or influential business figures who are crucial to maintaining state repression—the opposition is more likely to view these measures positively. Targeting these figures can reduce the risk of a rally around the flag effect since opposition groups may see the sanctions as a way to weaken the repressive apparatus that sustains authoritarian rule.

Thus, unlike the common categorization of sanctions simply as comprehensive or targeted in the literature, we distinguish between types of targeted sanctions based on whether they target centrist/reformist politicians or more hardline/die-hard entities. As discussed below, this approach helps us better explain the effects of comprehensive and targeted sanctions on different segments of the political spectrum and assess their varying impacts on public sentiment and opposition dynamics.

Similar to the winning coalition, citizens are also not homogeneous, as they have a range of attitudes toward the incumbent government. While existing literature on economic sanctions typically assumes that the general populace is the unitary actor in protests or rallying around the flag, recognizing the heterogeneity in the population provides a better understanding of why not everyone rallies around the flag (Edwards and Swenson 1997). Baum (2002) investigated all US major international events between 1953 and 1998 and showed that the propensity to rally around the flag varies according to individual and environmental circumstances. Building on this insight and aiming to explore who rallies around the flag in authoritarian countries in response to economic sanctions, we assume that citizens are heterogeneous regarding their attitude toward the status quo and possible methods of change. To align with the commonly discussed pro- and anti-government classifications, we first divide citizens into two main political groups in our model: loyalists and opposition factions, which are based on the citizens’ attitudes toward the incumbent government and the status quo. Loyalists and staunch supporters of an authoritarian state usually exhibit a high level of organization and unity, as the state has the motivation and resources to coordinate among them. However, the opposition lacks the same level of concentrated resources and motivation to dictate or enforce coordination among its members. In fact, anti-authoritarian opposition movements typically promote democratic values and encourage diverse opinions within their ranks. In other words, while organizational cohesion within a movement can contribute to its success, authoritarian efforts to coordinate can weaken and even dissolve coalitions across various opposition groups. Therefore, it is common for differences to be more tolerated within the opposition camp. Another dimension of heterogeneity within the opposition movement is their difference regarding the method of resistance. While some advocate for non-violent methods of resistance for normative or strategic reasons, others view violence as the only solution against a repressive authoritarian state (Chenoweth and Stephan 2012). Where the opposition campaign fails to change the status quo by either reforming or unseating the incumbent authoritarian regime, the choice of resistance strategies itself can become a major source of internal disagreement.

Thus, it is crucial to recognize the highlighted heterogeneity within opposition groups. Factors such as ideological stances, resistance strategies, and visions for a post-regime future can vary substantially among them. To capture these dynamics, we categorize the opposition into two groups based on their differing operational philosophies (DeNardo 1985). The radicals tend to prioritize their objectives, often showing less concern for the means by which they achieve them. Conversely, the moderates are more cautious, employing measured strategies and tactics. The collective impact of sanctions on unity or fragmentation is contingent upon the cumulative reactions of these diverse opposition factions and the wider public. For instance, moderate and nationalist opposition activists can view sanctions as breaches of national sovereignty, potentially further moderating their critiques of the regime or even intensifying their support. The visibility and sway of these moderate groups are pivotal in determining whether sanctions foster cohesion or division. Consequently, our analysis bifurcates the opposition into radical and moderate subgroups. Therefore, we identify three political groups in our model: radical opposition, moderate opposition, and loyalists. While we acknowledge that citizens’ attitudes toward the state can range from staunch loyalty to radical opposition, we focus on these three main political groups to facilitate the exploration of potential variations in the rally around the flag effect in response to different types of economic sanctions.²

The combination of three types of sanctions and three categories of citizens creates a 3 × 3 matrix, as shown in Table 1, for examining variations in the rally around the flag effect. The radical opposition has the largest ideological gap with the incumbent authoritarian state. They advocate for escalating tensions and are willing to use any tactics, including violence, to overthrow the regime. Unlike other opposition groups, the radical opposition considers no role for moderate members of the incumbent state in their transition plans. They believe that change can only occur through the complete dismantling of the current government, even if it involves violent means. Violence and its collateral damage are often justified by these radicals as an unfortunate but necessary sacrifice for the greater good of overthrowing an authoritarian state. Consequently, they view foreign interventions that weaken the targeted state and increase civilian grievances as beneficial. For this reason, the radical opposition is unlikely to rally around the flag in response to either comprehensive or targeted sanctions since they perceive the benefits of such sanctions outweigh their “justified” costs. Instead, they might even welcome these sanctions as they align with their goal of destabilizing the authoritarian regime. Thus, in Table 1, column 1, the radical opposition does not exhibit any support for rallying around the flag in response to any type of economic sanctions. We indicate this lack of rallying with a × sign.

Moderates, on the other hand, are more cautious about their strategies and tactics. They rely on civil resistance and, sometimes, reform from within to change the status quo. Consequently, they are careful when selecting their strategies and targets to maximize defections from the authoritarian side and mobilization of bystanders (Chenoweth and Stephan 2012; Chenoweth, Hocking, and Marks 2022). Because their approach is grounded in people power and aims to encourage defections among moderate members of the incumbent government, weakening the people and moderate forces within the regime can undermine their resistance strategy and hinder their efforts to bring about change. Therefore, they often view comprehensive sanctions skeptically, leading to potential rallies around the flag effects among them. Regarding targeted sanctions, moderates favor actions against die-hard figures and groups, which are crucial to the regime’s stability. However, from a strategic point of view, they are likely to withstand or even oppose sanctions targeting individuals or organizations that can be beneficial in driving democratic reforms and ensuring a peaceful transition to democracy. We use a ✓ sign in Table 1, column 2, to indicate rallying around the flag in response to comprehensive sanctions and targeted sanctions against centrist politicians. The absence of a rally-around-the-flag response to targeted sanctions against die-hard politicians is denoted with a × sign.

The final category in our proposed typology is the regime loyalists (refer to Table 1, column 3). Like the moderate opposition, loyalists are averse to comprehensive sanctions. Such measures not only disrupt the daily lives of the populace—including loyalists themselves as the main beneficiaries of the authoritarian state—but also pose threats to the regime’s stability and longevity, which they staunchly back. However, when it comes to targeted sanctions, their stance diverges from the moderate opposition. Sanctions aimed specifically at die-hard politicians and groups, potentially impacting them directly, are met with strong disapproval, often manifesting as a rally around the flag effect. Objections to targeting die-hard politicians may have purely ideological motivations, but they can also stem from economic reasons. As Lektzian and Patterson (2015) argue, economic sanctions create distinct winners and losers. Targeting die-hard politicians and organizations can negatively impact the economic benefits and rents that loyalists receive as compensation for their support of the regime. While one expected reaction might be that loyalists put pressure on the state to change its controversial policies, another more likely response, due to the risk of opposing an authoritarian state, is that they oppose the sanctions and increase their support for the state even further, hoping to compensate for the potential loss of benefits.

Whether driven by ideological or financial motives, targeting die-hard state entities can boost loyalists’ support for the incumbent government. Also, just as the moderate opposition might hesitate to support sanctions against centrist politicians and even oppose them, loyalists may see little reason to back those who seem less devoted to the regime. In addition, for loyalists, moderate politicians often represent competitors for political and financial benefits within the winning coalition. This competition is indeed institutionalized in competitive authoritarian regimes with the aim of increasing their survival chance (Gandhi and Heller 2018; Bernhard, Edgell, and Lindberg 2020). Therefore, economic sanctions that target moderate politicians and undermine their position generally do not result in rallying around the flag among the loyalists. Instead, such sanctions can even radicalize centrist politicians, driving them closer to the regime’s loyalist camp. In Table 1, column 3, the response to comprehensive sanctions and targeted sanctions against die-hard politicians is indicated with a ✓ sign, reflecting a rally-around-the-flag effect. The absence of this response to targeted sanctions against centrist politicians is represented with a × sign.

The above discussions can be summarized into three primary theoretical expectations:

Expectation 1: The radical opposition is unlikely to rally around the flag in response to any type of sanctions.

Expectation 2: The moderate opposition is likely to rally around the flag in response to comprehensive sanctions and targeted sanctions against centrist politicians but not in response to targeted sanctions against die-hard politicians or organizations.

Expectation 3: Regime loyalists are likely to rally around the flag in response to comprehensive sanctions and those targeted at die-hard politicians and organizations, but not when the sanctions target centrist politicians.

Research Design and Methods

We empirically investigated our theoretical expectations by studying the tweets of Iranian influencers on Twitter. Iran is an ideal case for this study for several reasons. First, Iran is not a democratic state; it can be classified as a competitive authoritarian regime (Levitsky and Way 2002). Although elections and democratic institutions are part of the political system, extensive rule violations, electoral manipulations, and restricted competition prevent it from meeting the minimal standards of a true democracy. This political environment allows us to examine differences among politicians, specifically centrists and die-hard factions, who compete within the Islamic Republic’s competitive authoritarian structure.

Second, the Islamic Republic of Iran, established in 1979, has faced opposition from various political groups almost from its inception. While these groups share a common opposition to the Islamic Republic, they adopt different resistance methods and envision different outcomes after the regime’s overthrow. Malek (2023) and Rivetti and Saeidi (2023) explore the fragmentation and genealogy of opposition movements in Iran. As these studies highlight, even during the recent “Woman, Life, Freedom” uprising—the most significant challenge to the Islamic regime since its establishment—the divisions and rivalries among anti-regime factions were evident.

Third, Iran has experienced a wide range of sanctions. Since the 1979 hostage crisis at the US Embassy in Tehran, the US government has imposed sanctions on Iran, increasing them as tensions grew between the Islamic Republic and the United States in the Middle East. As the Islamic regime’s nuclear program gained momentum in the early 2000s and nuclear talks failed in 2006, European countries joined the United States in imposing sanctions (Fayazmanesh 2008). In 2011, the United States and the EU coordinated to implement severe sanctions on Iran’s energy, finance, and trade sectors, effectively cutting Iran off from the SWIFT international banking system (Samore 2015). These sanctions were mostly lifted following the 2015 nuclear deal, the Joint Comprehensive Plan of Action (JCPOA). However, the United States withdrew from the JCPOA in 2018, reimposing sanctions as part of the Trump administration’s maximum pressure policy. This decision has led to further tensions between Iran and the United States in the Middle East, and the United States, in response, imposed sanctions on several Iranian political figures and entities, including Foreign Minister Javad Zarif and the Islamic Revolutionary Guard Corps (IRGC). The range and frequency of sanctions, which include both comprehensive and targeted types during the Trump administration, provide a unique opportunity to examine whether these various sanctions were associated with a rally around the flag effect among different Iranian political groups.

Iran’s competitive authoritarian framework, its diverse opposition groups, and the range of sanctions imposed on the country provide a nice setting for studying the dynamics of political competition and resistance in a non-democratic setting. Although Iran’s position in global politics might seem unique, the theoretical foundations of our proposed typology are commonly observed worldwide. The rise of authoritarianism and the opposition groups that form in response, along with the widespread use of sanctions as a relatively lower-cost method for addressing authoritarian states’ rogue and aggressive behavior, are significant trends in world politics nowadays. Therefore, studying Iran can provide valuable insights into these unfolding events in countries like Russia and China.

Our study relies heavily on data collected from Persian (Farsi) Twitter. Although Twitter may not perfectly represent Iranian public opinion, there are compelling reasons to use this platform for research in this context. First and foremost, our dataset comprises a diverse range of individuals, including politicians, political activists, journalists, and celebrities. These groups hold significant sway in shaping policies, influencing public opinion, and disseminating information through various channels. In fact, some of them are policymakers themselves, while others possess the ability to impact policies or sway public sentiment through their influential reach.

Second, conducting reliable surveys in non-democratic countries can be a formidable challenge. In such contexts, turning to online methods and platforms often emerges as the most viable option for studying the underlying micro-level dynamics that contribute to the macro-level outcomes we observe. This aligns with the insights discussed by Barberá and Rivero (2015), who discuss the challenges of utilizing data from social media platforms like Twitter but also discuss the importance of social media data use and interpretation.

Lastly, it is worth noting that there is an association between online and offline political activism (Best and Krueger 2005; Kim, Russo, and Amnå 2017; Greijdanus et al. 2020; Chayinska, Miranda, and González 2021). Recognizing this correlation, the utilization of Twitter data offers a unique opportunity to gain deeper insights into the dynamics of phenomena, such as the rally around the flag effect and its role in the (in)effectiveness of economic sanctions imposed on non-democratic countries. Thus, while Twitter may not provide a comprehensive study of Iranian public opinion, this platform, hosting socially influential and politically active citizens, does offer a valuable lens through which we can better explore and understand the interplay of politics, public sentiment, and policy outcomes in a complex and dynamic landscape.

The list of Iran-related influencers on Twitter was meticulously curated by Jalaeipour and Hajizadegan (2020). To qualify as an influencer on their list, three distinct criteria must be satisfied: (i) The account consistently tweets about Iran; (ii) it possesses a following of at least 10,000 users; and (iii) the account’s follower count is approximately double its number of followings. These criteria ensure that the selected accounts significantly influence discussion of Iran’s affairs on Twitter. This process identified 1,765 Iranian Twitter influencers. Additionally, Jalaeipour and Hajizadegan (2020) categorized the political affiliations of these Twitter influencers into six groups:

• Conservatives: vocal and loyal supporters of the regime.

• Reformists: advocates for reforming the Islamic Republic and its constitution within existing political structures.

• Transitionists: seek regime change through civil resistance.

• Overthrowers: aim to topple the regime by any means, even if it involves foreign military intervention.

• Non-politicals: predominantly athletes and artists.

• Unclear: mainly journalists who maintain neutrality in their tweets.

In alignment with our suggested typology presented in Table 1, we can categorize the Overthrowers as constituting the radical opposition. On the other hand, the Reformists and Transitionists fall under the classification of moderate opposition to the regime. Lastly, the Conservatives can be identified as the loyalists of the regime.

We gathered the tweets by connecting to the Twitter API using the usernames of these accounts. This yielded approximately 2 million tweets, posted between November 8, 2016, when Donald Trump was elected US President, and August 20, 2019, about 1 year after re-imposing the comprehensive sanctions. The Twitter API provides comprehensive tweet details, but we prioritized the text and the tweet’s creation date for our study.

Our primary metric revolved around gauging each influencer’s sentiment regarding the Iranian government. To achieve this, we conducted an analysis of tweet content, excluding replies. Within our extensive corpus, each tweet was categorized as either supportive (+1), neutral (0), or critical (−1) of the state. Given the substantial volume of tweets, we harnessed machine learning algorithms to ascertain their sentiment. This process entailed the manual annotation of a representative sample, followed by the optimization of algorithm parameters to enhance out-of-sample predictions.

The majority of tweets tended to exhibit a neutral or critical stance toward the Iranian government, resulting in a relatively sparse representation of supportive tweets. To mitigate this imbalance, we stratified tweets based on seven significant foreign crises (n_i = 800; i = 1,…, 8), including economic sanctions, thereby ensuring the robust training of our algorithms. Table 2 provides an overview of these seven foreign crises and their corresponding counts of manually coded tweets. Furthermore, to guard against potential bias, we manually coded 1,000 tweets unrelated to these events. Although the main aim of this study is to explore the rally around the flag effects in response to economic sanctions, we use all major foreign crises for stratified training of our machine learning algorithms. This is because foreign policy events often increase online discussions about supporting or opposing the regime, helping us to deal with the sparsity problem that we mentioned above. To this end, we use all the major foreign crises during the study period to train our algorithms for machine coding of attitudes toward the Iranian government, but our rally around the flag analysis, below in the empirical results section, primarily examines events related to economic sanctions. In the conclusion section, we consider the broader implications of our findings, discussing how responses to economic sanctions might inform our understanding of reactions to other types of foreign crises.

Therefore, to further balance the attitude classes, we up-sampled the 6,600 annotated tweets using Python’s Synthetic Minority Over-Sampling Technique function. Subsequently, these tweets underwent processing through a standard natural language processing (NLP) pipeline as follows:

1. Text cleansing.

2. Bag-of-words and TF-IDF vectorization.

3. Logistic Regression (outperforming both Naive Bayes and Random Forest—see Online Appendix 2 for model accuracy comparisons).

4. Parameter tuning via Cross-Validation Grid-Search (refer to Online Appendix 3).

Our model achieved an F1-Score exceeding 70 percent (see Online Appendix Table B.3). In a subsequent experiment, we employed a strategy of categorizing tweets based on the political affiliations of their authors, leading to the training of distinct classifiers for each affiliation category. This approach was grounded in the observation that individuals sharing the same political leanings often convey comparable sentiments, thus serving to mitigate noise and enhance classification accuracy. Our intuition was substantiated, particularly in the case of the Conservatives and Overthrowers groups.

Taking into account both methodologies, we applied the first method to classify tweets from Reformists, Transitionists, Non-Politicals, and Unknowns, whereas tweets from Conservatives and Overthrowers were categorized using affiliation-based algorithms. As a result, the F1-Score for each political affiliation exceeded 60 percent, contributing to an overall F1-Score that surpassed 70 percent. Accomplishing such precision, particularly given the dataset’s challenges, such as mixed-language content, diverse Persian linguistic styles, and controversial topics,³ holds considerable significance.

Among the seven events detailed in Table 2, three of them were characterized as economic sanctions. These three sanction events were of particular significance as they constituted our independent variables according to our proposed typology in 1.

The first instance involved President Trump’s decision to withdraw from the Iran Deal and the subsequent re-imposition of a substantial number of economic sanctions. Considering their wide-reaching economic and financial implications, these sanctions could be deemed comprehensive, impacting the entire population.

The second event centered around the imposition of sanctions by the United States on Iran’s Foreign Minister, Mohammad Javad Zarif. Zarif is considered a moderate politician with a notable educational background in the West, having graduated from the Josef Korbel School of International Studies at the University of Denver. His extensive tenure in the United States, where he represented the Islamic Republic at the United Nations, contributed to a reputation and the development of an influential network in both New York City and Washington, DC. Zarif is recognized for his ability to present a moderate picture of the Islamic Republic in the realm of global politics. It is essential to highlight that, in comparison to other ministers, he placed a greater emphasis on Iran’s broader aspects beyond the strictly Islamic facets of the Iranian government. This approach garnered him popularity among Reformists and, at times, among Transitionists, although it did not resonate well with Conservatives due to his moderate stance. This event can be regarded as an example of targeting centrist politicians.

The final event of relevance to our study was the designation of the IRGC as a foreign terrorist organization (FTO) by the US government. This action can be understood as targeting a staunchly committed and die-hard organization.

We use regression analysis to assess our developed theoretical expectations to study the association between Iranian Twitter influencers’ attitudes and the imposed economic sanctions. Our chosen unit of analysis is the account-day. Although our data give us the flexibility to aggregate data at finer temporal resolutions, such as hours and even minutes, we deliberately selected the day as the temporal aggregation unit. While such narrow temporal focuses enable us to observe changes in outcome variables over time with greater clarity, they can potentially introduce measurement errors and capture extraneous noise in our analysis. Therefore, we adopted a balanced level of temporal aggregation for our analysis. It is better than commonly used yearly analysis but is less liable to the above issues with a finer temporal aggregation.

Equation (1) shows our identification strategy for estimating the association between the imposed sanctions and influencers’ attitudes. Our dependent variable is the daily change in the average attitude toward the Iranian government. This allows us to compare each account’s attitude with its attitude in the previous day as a benchmark. The independent variables are the presence of sanctions imposed in the current period (Table 2), measured using dummy variables (E_i,t for i = 1,. . .,3). Specifically, if a given day coincides with one of the three aforementioned sanctions events, we denote it as E_i = 1; otherwise, it is assigned a value of zero.

We also include the average sentiment recorded from the previous day (S_i,t−1) as well as both time (month) and account fixed effects. This is important in mitigating the potential omitted variable bias that could influence our analysis and lead to a biased estimation. We employed the ordinary least square method to estimate these models:

Empirical Results

The changes in political group attitudes toward comprehensive and targeted sanctions are illustrated in Figures 1, 2, and 3.⁴ These figures capture changes in political groups’ attitudes toward the Iranian government in response to the United States withdrawing from the Iran Deal, imposing sanctions on Javad Zarif, and designating the IRGC as an FTO organization.

These bubble charts display the weekly average of each group’s attitude toward the Iranian government on the horizontal axis. The vertical axis represents the total likes received by these influencers during a specific week. The size of the bubble signifies the tweet volume per group for that week,⁵ and the bubble color indicates political affiliations. This setup provides a visualization of influencers’ average reactions to sanction events based on their political leanings.

A movement of bubbles toward the right indicates growing state support by influencers, while an upward shift suggests that this sentiment has gained more popularity (measured in likes) among the broader Twitter audience. Importantly, bubbles moving northeast during the sanction week indicate a “rally around the flag” sentiment among both influencers and the general audience. On the other hand, a movement toward the left and downward signals decreasing state support and increasing criticism on Twitter.

Figure 1 shows a predominantly negative sentiment toward the Iranian government among influencers in the week leading up to the United States’s Iran Deal withdrawal. Only Conservatives offered a positive sentiment. However, during the week of the withdrawal, represented in solid colors, there is a noticeable shift toward the right across all groups, indicating increased support for the Iranian government. Conservatives, Reformists, and Transitionists display the highest levels of support. Notably, even the Overthrowers demonstrate some level of support. However, from this visualization, it is not evident that the rally in support of the government among the Overthrowers is statistically significant. We will explore this further through regression analysis later in this section. The figure indicates that while this support diminishes somewhat in the following week, it remains higher than the initial levels for most groups.

Figure 2 displays the sentiments toward the Iranian government the week before, during, and after sanctions were imposed on Foreign Minister Javad Zarif. As seen in Figure 1, the sentiments are predominantly negative toward the Iranian government, except among Conservatives. The imposition of this targeted sanction leads to notable shifts in attitudes. Reformists and influencers with Unclear affiliations demonstrate a significant rally-around-the-flag effect. Conversely, Conservatives show less support for the regime, and Transitionists express increased criticism. Thus, targeted sanctions against a moderate centrist politician elicited varied reactions across different political groups. Theoretically, as anticipated, neither Conservatives nor Overthrowers displayed any rally around the flag effect; in fact, the former even decreased their support for the government. While we will delve deeper into these findings through regression analysis later, the visual representation of attitude shifts reveals that Transitionists, representing a moderate opposition group, did not rally around the flag, in contrast to the other moderate opposition group, the Reformists. An interesting observation is that the network of peripheral Twitter users demonstrated greater support for the Reformists’ rally around the flag response compared to Transitionists’ critical stance during these weeks. In fact, when the Transitionists intensified their criticisms against the government following the imposition of sanctions on Javad Zarif, a centrist politician, they received fewer likes on average. This indicates that the Transitionist influencers’ intensified criticisms of the state during the sanction week did not receive support from peripheral users. We will discuss this observation further in the conclusion section, along with its implications for the Iranian opposition movement.

Echoing our theoretical expectations, sentiment visualizations in Figure 3 reveal that designating the IRGC as an FTO—a targeted sanction—did not provoke the pronounced rally around the flag effect observed during the US withdrawal from the JCPOA.

However, as discussed in the theoretical section, a notable exception is seen among Conservatives, who are closely aligned with both the IRGC and the Iranian state and who displayed significant support for the state in their tweets. Although the figure shows a slight increase in support for the Iranian government among Reformists, our subsequent regression analysis reveals that these changes are not statistically significant.

The weekly visualization of changes in the sentiments toward the Iranian government, Figures 1–3, effectively conveys sentiment shifts in response to sanctions, mostly consistent with our theoretical expectations. However, we employ regression analysis to robustly examine our theoretical propositions.

This analyzes the impact of the three discussed sanction events on influencers’ attitudes toward the Iranian government, stratified by political alignment. Table 3 presents the model estimations, showing how different political factions reacted to comprehensive and targeted sanctions.

We visualize the estimated coefficients in Table 3 for each event to simplify cross-group sentiment comparisons. Figure 4 shows that, on average, the re-imposition of sanctions due to the Iran Deal withdrawal bolstered Iranian government support among influencers. However, a careful examination reveals heterogeneous reactions across groups: Conservatives, Reformists, and Transitionists all exhibited a positive change in their sentiment, though the latter’s response was not statistically significant (p_value > 0.1). Conversely, Overthrowers, Non-Political entities, and accounts with Unclear affiliations did not exhibit any statistically meaningful reactions.

The aftermath of the IRGC’s designation as an FTO is illustrated in Figure 5. This targeted sanction did not spark widespread government support. Conservatives, however, significantly rallied behind the state during this period. Interestingly, Transitionists increased their critiques, which can be targeting the IRGC—a foundational state institution in Iran.

The reactions to sanctions against Foreign Minister Zarif (Figure 6) show a different pattern. The full sample analysis shows a rally around the flag effect. However, when we focus on political groups, we find different results. The sentiment toward the government among both Reformists and Transitionists improved—although for the latter, this was not statistically significant (p_value > 0.1). Unaffiliated accounts also leaned more supportive, while this is not statistically significant as well (p_value > 0.1). Given the large sample size, the wide confidence intervals are attributable to substantial variance in attitudes within these groups, indicating significant levels of disagreement among members. While Javad Zarif was the Foreign Minister of the Islamic Republic, Conservatives did not rally around the flag, consistent with our theoretical discussions.

The results of our analysis not only corroborate our proposed typology and the ensuing theoretical expectations but also have significant implications. The magnitude and significance of the rally around the flag effect are influenced by the type of economic sanctions and differ among political groups. These findings underline the importance of studying the micro-foundation economic sanctions for studying their failure and success. This understanding is also crucial for policymakers, practitioners, activists, and individuals interested in international relations and political science, as it sheds light on the complex dynamics of political sentiment during times of crisis, specifically economic sanctions.

Discussions and Conclusion

This article explores the variations of the rally around the flag effect within a targeted authoritarian state, particularly among its opposition groups. Drawing on prior research, we developed a theoretical typology to investigate whether and how different types of economic sanctions influence this effect across diverse political factions. Previous studies have demonstrated that these non-violent foreign policy tools can exacerbate conflicts and heighten human rights violations in target states, as economic grievances often lead to protests that are brutally suppressed by authoritarian regimes. The effective and harsh repression of these protests is frequently argued as a reason for the failure of economic sanctions. Our research contributes to the scholarship on the failure of economic sanctions by showing that such measures can paradoxically bolster support for the incumbent government, even in non-democratic settings and among their oppositions.

These insights make valuable contributions to our understanding of the dynamics surrounding economic sanctions and lay the foundation for future research aimed at enhancing our understanding of this frequently employed but often unsuccessful foreign policy tool against authoritarian states. Our findings provide micro-level support for several theoretical arguments that were previously underexplored in social science research on economic sanctions. Notably, a discernible “rally around the flag” effect is observable at the individual level in response to comprehensive sanctions against an authoritarian state. Additionally, empirical evidence suggests that comprehensive sanctions produce a more substantial rallying effect. The magnitude of this effect for targeted sanctions is contingent on the targeted political actor, closely tied to nationalist sentiments and perceptions of an assault on national sovereignty and citizenry.

A pivotal discovery from our study is that the rally around the flag effect, spurred by economic sanctions, can itself cause discord among opposition groups. This insight is critical for understanding the dynamics of mobilization and the organizational cohesion among opposition factions in authoritarian states. While advocates of sanctions might argue that such measures weaken the state and incite new protests, they often fail to recognize that mere mobilization does not guarantee the successful implementation of policy changes in targeted authoritarian regimes. Effective opposition movements require not just public mobilization but the establishment of a robust coalition among key opposition groups (Bramsen 2018; Thurber 2019).

In this study, we focused on Iran and examined how influencers on Persian (Farsi) Twitter responded to three distinct episodes of comprehensive and targeted sanctions. RezaeeDaryakenari, Asadzae, and Thies (2024) show, using micro-level data from households, that economic sanctions have decreased the quality and quantity of food consumption significantly in Iran. Therefore, it is not surprising that the Islamic Republic has experienced several waves of protests over the last decade (Kozhanov 2022; Rivetti and Saeidi 2023). The regime has survived all these uprisings by relying on severe repression and thus becoming structurally more regressive (Davenport and Appel 2022).

While these sanctions-induced protests might appear as opportunities for change, they failed to unset the incumbent regime. Undoubtedly, state repression played a crucial role; however, deep divisions within the opposition campaign also hindered the formulation of a unified transition agenda. As Malek (2023) notes, despite efforts and occasional successes, the Iranian opposition faces a significant challenge in achieving an adequately cohesive and united front against the regime to successfully face its capable repressive apparatus. While this study does not delve into the reasons behind this fragmentation, our findings indicate that economic sanctions could exacerbate the situation. Both our descriptive and regression analyses revealed divisions among opposition groups in their reactions to economic sanctions. In fact, our results show that economic sanctions incited a rally around the flag effect in support of the Islamic regime, not only among its loyalists but also, in some instances, among opposition groups. This phenomenon may explain why uprisings like the 2019–2020 “Bloody November,”⁶ which was primarily rooted in economic grievances, did not garner significant support from the so-called “political elites” and ultimately failed despite approximately 1,500 deaths.⁷

Our findings suggest that economic sanctions created a rally around the flag effect among influencers on Twitter, a platform hosting diverse political elites and activists. Consequently, it is plausible to argue that political elites, even within opposition groups, exhibited different attitudes toward the government compared to the economically vulnerable citizens during popular uprisings, which followed economic sanctions. While the masses were frustrated with the economic burdens imposed by sanctions, elites and activists rallied around the flag. Therefore, the horizontal divide among opposition groups also translated into a vertical divide between elites-activists and the masses, complicating the prospects for successful regime change. These types of polarization, coupled with the higher level of human rights violations, can increase the risk of violence by resistance factions and diminish their likelihood of success (Pearlman 2011; Chenoweth and Stephan 2012; Pinckney and RezaeeDaryakenari 2022).

Above, we discussed the divide between elite-activists and the masses, yet this warrants a more systematic analysis. Using Twitter data prevented us from further exploring these divides. However, future research can entail a comprehensive study of both the horizontal polarization between political groups and the vertical polarization between these groups and the masses during foreign policy events. This will help us to determine whether and how the rally around the flag effect varies across less politically active populations, and how this potential divide between them and political activists in opposition groups may influence the success of opposition movements in achieving their goals.

We also should note that, despite general support for our theoretical expectations, there were instances where moderate political groups, namely Reformists and Transitionists, did not behave consistently. One avenue for future scholarship is refining our proposed typology to explain these observed patterns better. Applying the median voter theorem (Downs 1957) to opposition movements suggests that moderate opposition groups can play a crucial role in dominating discourses around shaping a unified movement agenda and reducing its fragmentation. Therefore, a better understanding of how moderate groups respond to economic sanctions can enhance our understanding of the factors that influence the effectiveness of these policies.

Moreover, as with any observational analysis, we cannot entirely eliminate various concurrent mechanisms that operate alongside economic sanctions. We employed a fixed-effects (FE) method to mitigate the risk of biased estimations due to omitted variable issues; however, other significant mechanisms warrant careful examination. For example, our case of comprehensive sanctions was not merely a re-imposition of comprehensive sanctions but also involved withdrawal from the Iran Deal, which was a crucial achievement for Reformists. We found evidence for a rally around the flag effect across different groups, so our analysis suggests that sanctions played a significant role and that the reputational cost of the JCPOA failure was not the sole reason for the rallying in support of the regime. Nevertheless, a more nuanced and systematic study to tease out these two parallel effects would enhance our understanding of how foreign policy events can affect support for authoritarian states.

Lastly, as discussed in the literature (Hale 2018), authoritarian states can exploit foreign policy crises to elicit patriotic sentiments even among their opposition. While this article primarily focuses on economic sanctions and investigates their failure by examining the rally around the flag effect, we employed seven different foreign policy events to train our machine learning algorithms. The bubble charts, Online AppendixFigures A.10–A.13, for the other four non-sanction events are included in the Online Appendix. These plots illustrate that various foreign policy events can also, to varying degrees, trigger a rally around the flag effect not only among Conservatives but also within some opposition groups. This rallying of support for authoritarian regimes is concerning as it can encourage such states to provoke additional foreign crises to bolster their survival. Therefore, another avenue for future research can involve theorizing and empirically exploring the micro-foundations of the rally around the flag effect in response to different types of foreign policy crises.

Author Biography

Babak RezaeeDaryakenari is a Senior (Tenured) Assistant Professor of international relations at the Institute of Political Science, Leiden University, with a research focus on political violence and conflict. On the methodology side, his primary focus is on applying machine-learning algorithms and social media data mining methods to studying and forecasting political conflict and violence.

Vahid Ghafouri is a PhD student in Telematics at IMDEA Networks Institute (+UC3M). He completed his BSc in industrial engineering at Sharif University of Technology and his MSc in business analytics at Sabanci Business School. His current main area of research involves the application of NLP and LLMs on social network data for measuring online polarization and radicalization.

Nihat Kasap works as a Professor of Business Analytics & MIS at the Sabanci Business School, Sabanci University, where he also served as the Dean between March 2018 and March 2024. His research focuses on social media analytics and data mining, mobile technologies and M-government applications, pricing and quality of service in telecommunication networks, generation expansion planning and investments in the energy sector, mathematical programming, and heuristic design and optimization.

Notes

Author’s note: We would like to thank the editors and anonymous reviewers for their valuable comments. The authors also extend their gratitude to Michael Colaresi, Lotem Halevy, Roseanne McManus, and the participants of the Amsterdam Conflict Research Network workshop, the Methodological Challenges of Studying Social Media workshop (funded by the Royal Netherlands Academy of Arts and Sciences-KNAW), the International Studies Association-Midwest 2020 Conference, and the Essex University Workshop for their insightful feedback. RezaeeDaryakenari is especially grateful for the support provided by the Gratama Stichting-Leiden University Fund. This project has been supported by the Gratama-Foundation and the Leiden University Fund (2020-06/W20330-5-GSL).

Footnotes

References