Populist Storytelling and Negative Affective Polarization: Social Media Evidence from Mexico

Abstract

The ideational definition of populism proposes that a narrative is populist if it is characterized by a Manichean cosmology that divides the political community between a “people,” conceived as a homogeneously virtuous entity, and an “elite,” conceived as a homogeneously corrupt entity. Departing from that conceptualization, this work first investigates the specific stories that Andrés Manuel López Obrador uses to spread his populist worldview, which we call “storytelling.” We define the idea of storytelling as the art of telling a story where emotions, characters and other details are applied in order to promote a particular point of view or set of values. Second, we explore whether some of those stories produce greater negative affective polarization, here defined as the extent to which rival sociopolitical camps view each other as a disliked out-group. Findings suggest that some specific stories—in particular, what we call “stories of conspiracy” and “stories of ostracism”—indeed tend to induce more polarized attitudes among citizens.

Populism has been defined by the ideational approach as a “set of ideas” (Hawkins 2009, 1045; Neumann-Ernst 2019; Priester 2012, 1; Roouduijn 2014, 3) that conceives politics as a Manichaean relationship between two homogeneous entities, a “pure people” and a “corrupt elite” (e.g., Hawkins 2009; Hawkins et al. 2019; Mudde 2004; Mudde and Rovira Kaltwasser 2013). According to this approach, populism is a narrative in which “[t]he people are opposed to what is frequently subsumed under the label of ‘the elite’” (Mudde 2004, 543). Thus, populism consists of a narrative with two components: people-centrism and anti-elitism (e.g., Mudde 2004; Hawkins et al. 2019; Rooduijn 2013, 6).

However, this same ideational approach has also described the populism as a narrative with “an identifiable but restricted morphology that relies on a small number of core concepts whose meaning is context dependent” (Mudde and Kaltwasser 2013, 150–151; Nordensvard and Ketola 2022, 862). Under populism, the composition of both “the people” and “the elite” can take a variety of different forms (Caiani and Graziano 2019). Consequently, the construction of a populist subject is “a context-specific product of political conflict, mobilization and discourse” in any given case (Roberts 2022, 683). In light of this, we herein propose that the abstract or vague nature of the two central concepts of populist narrative (i.e., “people” and “elite”) makes their meanings especially sensitive to the context in which they are used. This in turn makes such meanings theoretically relevant, because populist leaders intentionally use the ambiguous nature of those terms to fill them with different stories that make sense in a specific context. Thus, this article is interested in the stories that populist leaders put forward using the fuzzy notions of “people” and “elite”—and shows a particular interest in the effects of such stories on negative affective polarization among citizens.

The concept of affective polarization is derived from the classic notion of social distance (Bogardus 1947), and it “requires not only positive sentiment for one’s own group, but also negative sentiment toward those identifying with opposing groups” (Iyengar et al. 2012, 406). Particularly, this work is concerned with what we call “negative affective polarization”—following Samuels and Zucco’s (2018) idea on “negative partisanship” in their analysis of the polarizing opposition to the Workers’ Party (PT) in Brazil (see also Samuels et al. 2023). In this vein, we define negative affective polarization as the extent to which partisans or rival sociopolitical camps view each other as a disliked out-group (e.g., Iyengar et al. 2019). In other words, negative affective polarization refers to one of the two dimensions of the most used concept of affective polarization and it occurs when partisans of one side dislike and distrust those from the other (Iyengar et al. 2019). Assuming the absence of a positive sentiment towards one’s own group along the pro-AMLO versus anti-AMLO dimension, our analysis focuses on examining negative sentiment towards those who identify with the opposing group.

Hence, taking refuge in the ambiguous and abstract character of notions such as “people” and “elite,” populist leaders appeal to “narrative patterns, myth-making, and political emotions” in order to make populist ideas make sense to voters and thus obtain their political support (Nordensvard and Ketola 2022, 862; see also Ungureanu and Popartan 2020, 41). Thus, “[o]ne way the ‘emptiness’ of these signifiers garners meaning is through the stories they relate to” (Nordensvard and Ketola 2022, 862). Identifying the specific stories used by populist leaders is relevant to understanding the effects of their narratives on political attitudes among citizens in a particular context.

Departing from these ideas, this work aims to explore the specific stories that a populist leader, Andrés Manuel López Obrador (AMLO) uses to communicate with voters; in line with the current literature on political narrative (e.g., Polletta et al. 2011; Seargeant 2020; Stenmark 2015; Ungureanu and Popartan 2020), we call this communication strategy storytelling, defined as the “art of telling a story” with a “plotline containing emotions, agency, antagonism, heroes and enemies,” and which aims “to promote a particular point of view or set of values” (Nordensvard and Ketola 2022, 861–863). This is a characteristic of political narrative in general; in the specifically populist storytelling case, however, employing empty or vague signifiers such as “people” and “elite” frames a narrative that can “bind heterogeneous demands together” (Ungureanu and Popartan 2020, 40). Populist storytelling simplifies many areas of policy, which are themselves inherently complex and contested, into a few seemingly simple and incontrovertible stories.

Thus, the research problem of this article addresses the link between AMLO’s different stories and negative affective polarization among his followers in social media. In other words, the research question of this study is: Do AMLO’s different stories influence the levels of negative affective polarization on social media? We seek first to identify the particular stories that López Obrador uses on Twitter (recently rebranded as “X”), and second, we seek to identify which of these stories produces a greater number of polarized replies.

In this regard, the purpose of this work is purely exploratory. Our article investigates the relationship between populist storytelling and negative affective polarization using social media data. Thus, we are not testing the theory or conducting an explanatory case study. Our research design does not allow for a proper test of a causal argument. However, we argue that the value of the findings in this study, based on observational data, lies in two aspects. First, it provides an initial overview of the empirical links between populist storytelling and negative affective polarization. Second, it lays the groundwork for designing experiments in future research that can test the causal relationship suggested by the results of this study. Then, the contribution of this work consists of conducting a theory-generating exercise.

Therefore, this work seeks, using the case of the president of Mexico, Andrés Manuel López Obrador—who, according to the ideational approach to populism, seems to have exhibited a populist narrative (Monsiváis Carrillo 2022; Sarsfield 2024)—to explore the nexus between populist storytelling and negative affective polarization. To do so, this work uses a combination of human-based and automated-based textual analysis and both quantitative and qualitative textual analysis of social media, specifically Twitter. Findings suggest that: i) AMLO in his tweets employs eight stories to refer to “the people” and “the elite”; ii) replies by followers of the AMLO account on Twitter are more polarized when AMLO’s posts propagate what we call “stories of conspiracy” and “stories of ostracism”; and iii) polarized replies to different AMLO stories on Twitter are neither uniform nor homogenous but rather very diverse.

Storytelling and Populism

A more far-reaching understanding of populism must focus on the study of the storytelling that populist leaders use to appeal to both their electorate and to a broader base (Aslanidis 2016; De Cleen and Stavrakakis 2017; Halikiopoulou 2019; Moffitt 2016; Taş 2020, 218). Populist politicians “tell stories to make claims, enlist support, and defuse opposition” (Taş 2020, 140). Storytelling, in the case of populist leaders, contributes to building up a perceived body of evidence for and lends more public credibility to the thin ideology of populism focused on the Manichean opposition between a “pure people” and a “corrupt elite.” Because it is important that the nexus between populist storytelling and voters’ everyday experiences “does not become too artificial” (Nordensvard and Ketola 2022, 865), populist leaders intentionally employ some stories that reflect the life of common people.

Roughly speaking, storytelling entails narratives that “simplify complexity, selectively appropriating characters and events” (Nordensvard and Ketola 2022, 864), which are presented as causally and temporally related to each other (Ewick and Silbey 1995). There is evidence in the psychological literature showing that stories are a central device to help individuals make sense of their social and political world (Bruner 1991; Hase 2021, 686; Koschorke 2018; McAdams 2011; Polkinghorne 1988). Storytelling is, in this sense, a “sensemaking tool” that “does not simply consist in adding episodes to one another,” but “also constructs meaningful totalities out of scattered events” (Alvesson and Sköldberg 2018, 128). Thus, “[a]cts of storytelling are therefore highly effective in making sense of complex events and facts” (Nordensvard and Ketola 2022, 865), giving certainty to individuals facing the ambiguous and unclear reality of politics. Acts of storytelling are powerful because they help individuals to live with ambiguity (Stenmark 2015, 931).

In addition, storytelling exerts a cognitive effect by countering uncertainty using an intelligible, plausible story through which individuals make sense of the complexities of the social, economic, and political realms of the world. In particular, storytelling about policies influences voters because such stories create cognitive shortcuts by which citizens can address the uncertainty and ambiguity inherent to politics. What characterizes populist storytelling specifically is that by its selective recounting of past events and characters, it constructs a sequential order that helps audiences make sense of given events not primarily in causal terms, but in moral terms (Monroe 1996; Taş 2020). Populist storytelling is not mainly about facts, but about drawing moral distinctions. Even more, populist storytelling as a kind of narrative is, to a certain degree, independent of events (Hase 2021, 786; Koschorke 2018: 7–9, 202). Although populist storytelling draws on factual events, it is predominantly a “moral story with a clear sense of right and wrong, where the actors are located on one side or the other” of different political issues (Nordensvard and Ketola 2022, 874). Thus, by this power of moral “meaning-making and simplification … contemporary populist leaders gain adherents” (Taş 2020, 130).

Populist storytelling appeals not only to moral absolutes, but to emotion. Populist storytelling seeks to connect citizens with highly complex policy problems in an emotional and affective way, aiming to produce political engagement with populist leaders which otherwise would not have occurred (Rico et al. 2017; Salmela and von Scheve 2017). Thus, the underlying affective drivers of populist storytelling demonstrate why such storytelling is so powerful (Skonieczny 2018). By activating strong emotions such as anger and fear, populist storytelling challenges, erodes, and undermines the extant cognitive and normative frames with which citizens understand events in politics (Bronk and Jacoby 2020).

Given the inextricable nexus between emotions, morality, and stories that characterizes populist storytelling, it is fair to say that populism is performative. Populist stories represent performative acts (Laclau 2005) which are creators of meaning for the political word. Populist storytelling is “something performative” that populist leaders “use to engage” their audience “with a contested issue” (Nordensvard and Ketola 2022, 866). Stories “are more likely to be (or become) dominant if they are told by narrators with wide discursive reach” (Carstensen and Schmidt 2016, 326–328; see also Hase 2021, 687; Koschorke 2018, 198). Populist leaders, aware of the argumentative force of performative acts, appeal to the theatricalization of their ideas. The seductive quality of populist ideas requires a performative and credible populist narrator.

Conceptualizing and Explaining Negative Affective Polarization

There are various definitions of polarization in political science. Recently, there has been a growing debate concerning the meaning of the concept due to increased research in this area. With polarization processes expanding globally and the phenomenon having multiple dimensions (Roberts 2022), there has been a proliferation of alternative forms or types of the concept, including several examples of polarization “with adjectives” (Sarsfield et al. 2024, this special issue), such as social polarization (McCoy and Rahman 2016), populist polarization (Enyedi 2016), affective polarization (Iyengar et al. 2012), and pernicious polarization (McCoy and Somer 2018). The lack of consensus and the multidimensional nature of polarization become evident when these different definitions of the concept are examined.

Within this complex conceptual debate, the notion of affective polarization, along with that of ideological polarization, is key to the conceptualizations of polarization in the literature. In brief, while the concept of ideological polarization seeks to evoke spatial distance in terms of political preferences between parties or voters (e.g., Abramowitz and Saunders 2008; Roberts 2022), the notion of affective polarization refers to having positive feelings towards one’s own group and harboring negative sentiments towards individuals who identify with opposing groups (Iyengar et al. 2012, 406). Thus, negative affective polarization—the type of polarization that concerns us here—is defined in this work as the extent to which political or social groups view each other as a disliked out-group (e.g., Iyengar et al. 2019; Iyengar et al. 2012; Druckman and Levendusky 2019; Levendusky 2009). This definition of negative affective polarization is based on the degree of mutual animosity between opposing groups or “negative sentiment toward those identifying with opposing groups” (Iyengar et al. 2012, 406), to the point where one group may not recognize the other as legitimate actors in the democratic arena (McCoy and Rahman 2016). Specifically, this study is interested in the hostility between groups in the pro-AMLO versus anti-AMLO dimension (i.e., negative affective polarization between these two opposing groups).

When examining the theory of affective polarization (i.e., affective polarization as a dependent variable), the literature can be divided into two broad approaches. One group of theories emphasizes the impact of external factors in the generation of affective polarization, such as media, ideology and social sorting, elite influence, and political institutions. The other group of theories focuses on internal causes, such as social identity, stereotypes, motivated reasoning, and traits (Druckman and Levy 2022). Within that general context, a prominent theory of affective polarization argues that partisanship, as a social identity, underlies affective polarization, with changes to the media environment and social sorting amplifying this polarization. In this vein, in this work we are interested in the hypothesis that posits the changing media environment, specifically the different narratives that leaders use in their social media accounts, as one of the external causes of affective polarization.

Following the findings of a group of works, our article investigates the idea that certain narratives among elites play a significant role in driving negative affective polarization among the masses (Druckman et al. 2021). Affective attitudes in the public can be directly influenced by the tone of political elites, as shown by Gentzkow et al. (2019). Lau et al. (2017) conducted a dynamic process tracing experiment that demonstrates how diverse media environments and negative campaign rhetoric result in higher levels of affective polarization, especially negative affective polarization. From this approach, we specifically focus here on what we refer to as “populist storytelling” presented by these elites on social media who contribute to this phenomenon. Our results appear to align with those previous studies, emphasizing the impact of both the tone of political elites and negative campaign rhetoric on negative affective polarization.

Hypotheses: Expectations

Although this article is not intended to generate theory, since our work is exploratory and does not provide a strong foundation for testing causal theories, we briefly present our expectations here, proposing the potential theoretical and empirical connection between populist storytelling and negative affective polarization. It should be noted that testing these expectations as causal hypotheses would necessitate experimental data, which is not utilized in this study. Nevertheless, we belief that it is crucial to suggest our theoretical expectations and their empirical implications, at the very least.

Regarding the theoretical nexus between populist storytelling and negative affective polarization, the psychological concept of motivated reasoning holds that emotions influence the way individuals interpret information (Flynn et al. 2017; Taber and Lodge 2006). Given this, populist storytelling could trigger emotions that would tend to harden and sharpen the attitudes of voters on each side in ways that would make them not only more affectively distant but also less willing to cooperate or even coexist with each other (Carlin et al. 2019, 430).

Simplistic stories “based on defining individuals or societies in terms of a unique affiliation,” such as those used in populist storytelling, “can be used to foster strong feelings of within-group solidarity but also ones of between-group disagreement” (Cárdenas 2013, 789). It is the hostility between different groups that constitutes the phenomenon of interest in this work, that is, negative affective polarization. Following this idea, the research question of this article is: What is the relationship between populist storytelling and negative affective polarization in social media? In other words, do different stories in AMLO’s posts co-occur with differences in the levels or types of polarized replies on Twitter?

Hypotheses derived from the theory of affective polarization have been tested through case study and large- and small-n observational, longitudinal, and experimental research designs. Within these broad design families, this article adopts a multimethod approach (Brady and Collier 2010), using a combination of machine-based quantitative textual analysis and human-based qualitative textual analysis. Hence, this article’s methodology aligns with that of authors who favor measuring narrative using the traditional method of content analysis that decomposes the text and measures the components of ideas (e.g., March 2019; Grbeša and Šalaj 2019), which in this case are populist storytelling and polarization.

The expectations of this work are as follows:

• E1: Certain AMLO stories on social media lead to a greater negative affective polarization in the replies among the followers of his account;

• E2: Different AMLO stories give rise to polarized replies on Twitter among the followers of his account that are not uniform nor homogenous but rather very diverse, which suggests the generation of different dimensions or types of negative affective polarization according to different AMLO stories.

Measuring Populist Storytelling and Negative Affective Polarization

To measure our variables, we first measured AMLO’s populist storytelling through the tweets that López Obrador posts on his Twitter account. Second, we measured negative affective polarization through analysis of the replies that the followers of the AMLO account upload to such tweets. ¹ The analysis of AMLO’s tweets began with the selection of a set of words that are semantically and contextually equivalent to the attributes of populist narrative proposed in the ideational definition. First, given the conceptualization of populism that defines a narrative as populist if it is characterized by a cosmology that divides the political community between a “people” and an “elite” (Hawkins and Rovira Kaltwasser 2019; Mudde 2004), we selected those tweets with the contextually equivalent words “people” and “conservatives,” respectively. ² To represent the idea of the Manichaean character of politics (i.e., that “corrupt elites” benefit from “pure people”), we further selected tweets with the contextually equivalent words “corruption” and “privileges.”

The data collection covers all tweets from the creation of the AMLO account (2014) to the present (2022) (n = 3,249), plus all the replies to those tweets during the same period (n = 1,281,095). Within this population, we work with an intentional sample of those tweets that contains one of the words mentioned above, that is, “people,” “conservative,” “corruption,” and “privileges” (n = 302), and with the sum total of all replies to those tweets (n = 59,561). Given the research question of the article (i.e., which AMLO story produces more negative affective polarization), the sample selection procedure was purposeful and, therefore, non-probabilistic. A probabilistic sampling would have included AMLO posts irrelevant to answering the research question of this work—that is, posts that did not contain any of the theoretically relevant keywords and thus the stories that are narrated using these words—which consequently would make it impossible to answer the question about whether any of these stories produces negative affective polarization and, if so, which ones do so most strongly. ³

Regarding the data analysis, we proceeded as follows. In order to identify the stories that AMLO uses in his tweets (our independent variable), we used both quantitative textual analysis by machine learning (Natural Language Processing) as well as human-based qualitative textual analysis. Thus, we identified the terms that López Obrador uses most frequently in his tweets (i.e., a word cloud), the words most frequently associated with each other (i.e., analysis of co-occurrence), and the semantic relationships behind the co-occurrence of word pairings (i.e., qualitative textual analysis). From this data analysis we identified eight stories employed by AMLO.

To explore the level of negative affective polarization expressed in the replies that followers of the López Obrador account uploaded to AMLO’s tweets (our dependent variable), we further measure the percentage of polarized responses to each of AMLO’s stories. To do this, we tagged each of AMLO’s tweets to one of the eight stories previously identified (or to none of them, if this is the case) and measured the level of negative affective polarization of the replies to each of the eight stories as an average of the proportion of polarized responses that each tweet has within of each story. Also, we measure the level of negative affective polarization in the total number of replies and compare it with the number of polarized replies for each one of AMLO’s stories.

As mentioned earlier, following the conceptualization of negative affective polarization as the extent to which partisans or rival sociopolitical camps view each other as a disliked out-group (e.g., Iyengar et al. 2019)—in other words, as an “us versus them” view of the political world (McCoy and Rahman 2016)—we measured it in the pro-AMLO versus anti-AMLO dimension. We considered messages attacking the out-group or expressing intense hatred towards partisans of the opposing group as cases of negative affective polarization. In this vein, we made an assumption that there is no in-group affinity along the AMLO-supporters versus AMLO-opponents dimension. ⁴ Consequently, we focus on what we called negative affective polarization, a concept we coined to describe disdain for the out-group in the absence of positive affect among the in-group.

To measure polarized replies to the AMLO’s posts, we utilized a holistic grading schema (Hawkins 2009; Hawkins and Castanho Silva 2019). Replies to AMLO’s stories were coded as either “not polarized” or “polarized,” based on the following scale: 0) a reply in this category exhibits minimal if any negative polarized elements; 1) a reply in this category is opposed, in a hard tone, to the out-group, and specifically has few elements that could be considered nonpolarized.

AMLO’s Words

Exploring the words that AMLO uses the most in his tweets, we find some patterns that begin to outline the narrative structure of López Obrador. First, in AMLO’s tweets the terms “the people,” on the one hand, and of “mafia,” on the other are very frequently mentioned—the latter being a term by which, together with “conservatives,” AMLO alludes to a “corrupt elite,” as we shall see—which is evidence of the populist character of his narrative. Second, AMLO also widely uses the notions “Mexico,” and “country,” suggesting the occurrence of what has been called for other populist leaders as a specifically nationalist populism (Taş 2020).

Third, the high frequency of the terms corrupt, corruption, PAN (Partido Acción Nacional), PRI (Partido Revolucionario Institucional), Calderón (i.e., the former president Felipe Calderón), Salinas (i.e., the former president Carlos Salinas de Gortari), EPN (i.e., the former president Enrique Peña Nieto), and PRIAN—an acronym coined by AMLO to unite the names of the two parties that governed Mexico before him, PAN and PRI—stand out (Figure 1).

Figure 1. Word Cloud of AMLO’s Tweets, 2018

AMLO also very frequently uses the word MORENA (Movimiento de Regeneración Nacional), the populist party that AMLO leads. As we shall see later, the co-occurrence analysis of the tweets in which those words appear together suggests the use of a story affirming that “the people” are exclusively represented by AMLO’s party and that all opposition parties are equally “corrupt,” “conservative,” and part of a “mafia of power.”

Fourth, a characteristic that seems to be present in the words that AMLO uses the most is the so-called “chronopolitics … or ‘politics of time’, that is, how politics is about time” (Maier 1987, 151), and how “time [is] presupposed by politics” (Taş 2020, 128). In this vein, populists’ stories seem to “share a common narrative template to recount and connect the past, present, and future” (Taş 2020, 128), where the distant past is a utopia that was disrupted by some crisis in the recent past, and in front of which the populist leader presents himself as the savior and restorer of the distant and lost utopia.

In this vein, AMLO’s words exhibit, on the one hand, negative allusions to the recent past through a systematic and pejorative mention of the former presidents, candidates, and parties that were in power immediately before his government, which AMLO calls the ancien régime, and whom López Obrador considers as a homogeneously corrupt entity.

On the other hand, in the words that AMLO uses the most there are repeated allusions to Benito Juárez, Miguel Hidalgo, and José María Morelos (see Figure 2), among other national heroes, which portray López Obrador’s utopian vision regarding the distant past, a glorious era brought to an end by the neoliberalism of the recent past.

Figure 2. Word Cloud of AMLO’s Tweets, 2022

AMLO’s Stories

The analysis of the stories that AMLO uses begin, first and as mentioned above, with the selection of a set of words that are contextually equivalent to those attributes of populism that the ideational definition proposes. These words are people, conservatives, corruption, and privileges. Second, we identify the terms most associated in López Obrador’s tweets with those words. Then, we measure the co-occurrence of all these words by counting how many tweets they both appear in.

Empirical analysis encompassed both quantitative and machine-based tools in the study of the co-occurrence between words, along with qualitative and human-based textual analysis of the semantic relationships among such words by considering the co-text that accompanied, which allowed for identification of the stories that AMLO uses. In other words, in order to identify the stories used by López Obrador, we also employed qualitative textual analysis of AMLO’s tweets, because co-occurrence of words requires the analysis of the semantic meaning of such relationships. From this analysis emerged a set of different stories that AMLO narrates in a systematic way, and whose totality and structure we call “storytelling.”

Figure 3 shows the co-occurrence between the contextual terms of AMLO’s populist narrative (i.e., people, conservatives, corruption, and privileges: blue nodes) and the words most associated with such terms (orange nodes). A larger node indicates a higher frequency of the word. The thickness of the lines shows the strength of the co-occurrence between such words (thicker lines indicate higher co-occurrence). The color of the lines represents each of the stories we found in AMLO’s narrative.

Figure 3. AMLO’s Storytelling

Findings show the presence of a group of eight stories behind these co-occurrences, some of them similar in character to those found in the narratives of other populist leaders (e.g., Engesser et al. 2017; Nordensvard and Ketola 2022; Taş 2020). Following the comparative literature—although identifying some peculiarities of AMLO’s narrative—in this work we call these stories as follows: “Advocating for the people” (blue); “Democracy as direct democracy” (light green); “Corrupt and exploitative elites” (red); “Ostracizing the others” (gray); “Conspiracy theory” (purple); “Invoking the ‘heartland’” (light blue); “Executive aggrandizement” (green); and “Personal action frames” (orange) (see Figure 3).

Then, we first present a table (Table 1) providing a concise description of each populist story identified by comparative literature. Second, we offer a more comprehensive description of those stories, along with empirical evidence, as they relate to AMLO’s narrative.

Table 1. Main Populist Stories

Advocating for the People

One of the most robust findings of this work shows a very strong co-occurrence between people and well-being (Figure 3, blue). Likewise, and also around the word people, there is an important co-occurrence between people, justice, support for the people, and respect for the people. ⁵ These words form a network of terms that stem from the idea of people, which is the central node of this group of words. Thus, in the tweets in which AMLO uses the word people, he systematically mentions well-being, justice, support for, and respect for as well, forming a network of notions that suggest a prescriptive story that, following the literature, we call “advocating for the people.” As has been pointed out in other examples of populist rhetoric, in AMLO’s storytelling the “people’s needs and demands” are placed “above everything else” in politics (Engesser et al. 2017, 1112; Taggart 2000).

Qualitative textual analysis of the tweets confirms the idea that advocacy for the people is a key story of AMLO’s storytelling. Thus, AMLO insists in several of his tweets that the fundamental obligation of the government is to guarantee the rights and well-being of the people. A large number of tweets confirm the prevalence of this story, in which the “act of advocacy” is “performed by stressing” that AMLO’s (populist) government, unlike mainstream politicians, “is a true representative of the people” (Neumann-Ernst 2019, 104). Some examples are: “The government is at the service of the people” (El gobierno está al servicio del pueblo: https://t.co/KgVojJO7XP); “The most important thing for us is the welfare of the people” (Lo más importante para nosotros es el bienestar del pueblo: https://t.co/iIj5SvzhX2); “The government is now at the service of the people” (El gobierno está ahora al servicio del pueblo: https://t.co/MDI1N6kfiN); and “Now the welfare of the people comes first; nothing will make us go back to the past” (Ahora el bienestar del pueblo es primero; nada nos hará regresar al pasado: https://t.co/YAymsC9ybg).

Democracy As Direct Democracy. Emphasizing the Sovereignty of the People

A second story identified in AMLO’s storytelling poses the centrality of the people’s will and the absolute sovereignty of the people (Abts and Rummens 2007; Albertazzi and McDonnell 2008; Mudde 2004; Neumann-Ernst 2019; Shils 1956). Findings show a strong co-occurrence between the words democracy and people (Figure 3, light green). The demand for unrestricted popular power embodied in this story puts the role of democratic institutions—long characterized as one of the attributes that distinguishes liberal democracy (Abts and Rummens 2007; Dahl 1956)—in a distant second place. In this story, “elites are accused of having deprived the people of this right, rendering sovereignty the central subject of all subsequent disputes” and, therefore, only the leaders who truly represent the people (i.e., AMLO himself) are “able to restore the sovereignty of the people by replacing the elite and all other representative and intermediary institutions” (Engesser et al. 2017, 1111).

From this prescriptive idea in favor of emphasizing the sovereignty of the people emerges the conception from which AMLO links democracy with direct democracy. The notion of democracy in AMLO’s storytelling is associated with direct democracy, not with representative democracy. The story posits that because traditional politicians are corrupt, what is required is the participation of “the people.” The qualitative analysis of a large number of tweets finds AMLO’s arguments in conflict with the idea of representative democracy as a form of government: “The people have, at all times, the right to modify the form of their government” (El pueblo tiene, en todo momento, el derecho de modificar la forma de su gobierno: https://t.co/AdDFc8ANQR); “In democracy it is the people who rule” (En la democracia es el pueblo el que manda: https://t.co/VMFbWQKY4w); “Now the people are the main protagonist of this story” (Ahora el pueblo es el protagonista principal de esta historia: https://t.co/KVZjaoloQ1). The central actor that defines democracy is the “people,” not democratic institutions. In other words, it is the people, and not their representatives, who must govern.

Corrupt and Exploitative Elites

A third story identified in AMLO’s storytelling is directly related to one of the attributes that, according to the ideational approach, defines populism: the idea that “the elites” are corrupt and exploitative of “the people” (e.g., Albertazzi and McDonnell 2008; Hawkins et al. 2019; Mudde 2004). This story by AMLO ties the idea of corrupt and exploitative elites to those who had illegitimate “privileges” under the previous status quo. Results show that the word privileges is the central node of a network of terms such as corruption, ancien régime, unfairness, and legal privileges (Figure 3, light red), indicating as unfairly privileged all those who illegitimately benefited at the expense of “the people” in the recent past.

The qualitative textual analysis of AMLO’s tweets shed light on the semantic content of this story. In one of his tweets, AMLO affirms: “Those who felt they owned Mexico think that they will return for their rights and privileges” (Los que se sentían dueños de México piensan que regresarán por sus fueros y privilegios: https://t.co/E98BRE4U2q). Hence, in this story the elites are associated with privileges, injustice, and corruption supposedly prevalent during the ancien régime: “The Fourth Reform will uproot the corrupt regime of injustices and privileges” ⁶ (La Cuarta Transformación arrancará de raíz al régimen corrupto de injusticias y privilegios: https://t.co/zITA4wgaUJ). Faced with these unjustified privileges of the corrupt elites, the populist leader will give lost justice back to the people: “The privileges of the little pharaohs of the PRIAN will end soon” (Pronto se terminará con los privilegios de los pequeños faraones del PRIAN: https://t.co/XsT6GER1fv).

Ostracizing the Others

Another story that emerges from the analysis of AMLO’s storytelling is a morally negative characterization of “the others.” Research has found that in other populist leaders’ storytelling, “dangerous others,” operating separately from but with the acquiescence the elite, are identified as antagonistic to the people (Albertazzi and McDonnell 2008, 3; Engesser et al. 2017, 1112; Neumann-Ernst 2019, 99; Rooduijn 2014, 2). In this story, “[w]hereas the elite are considered as a danger from above (vertical dimension), the others are perceived as a threat from outside or within the people (horizontal dimension)” (Jagers and Walgrave 2007, 324).

Consequently, “the others are not regarded as part of the elite but as unjustly favored by the elite or even as their partner in a conspiracy against the people” (Engesser et al. 2017, 1112). In this story, the “dangerous others” have common cause with the elite in that they both oppose to “the people.” AMLO applies “the dangerous others” to journalists, intellectuals, and anyone who is critical to his government, associating them with “corruption,” “hypocrisy,” and to what AMLO calls the “mafia of power,” as shown in the co-occurrences between these words in Figure 3 (gray).

Qualitative textual analysis of López Obrador’s tweets confirms this hypothesis. For instance, AMLO addresses Professor Jesús Silva-Herzog to tell him: “Jesús Silva-Herzog Márquez has been questioning me with conjectures of all kinds for a long time. Today, in the Reforma news article, he unfairly accuses me of being an opportunist. No way, these are times to face the mafia of power, its henchmen and conservative journalists with the appearance of liberals” (Hace tiempo que Jesús Silva-Herzog Márquez me cuestiona con conjeturas de toda índole. Hoy, en el periódico Reforma, me acusa sin motivo de oportunista. Ni modo, son tiempos de enfrentar a la mafia del poder, a sus secuaces y articulistas conservadores con apariencia de liberales: https://twitter.com/lopezobrador/status/960607411315838977?lang=en).

What characterizes this story is that everything other than AMLO and his political supporters—whom AMLO encompasses all as conservatives—is homogeneously connoted by López Obrador with negative attributes such as corruption, hypocrisy, and mafia. In the story “ostracizing the others,” AMLO seeks to isolate “the others” by appealing to the moral sensibilities of his audience, defining “others” or “conservatives” as all who think differently from him.

Conspiracy Theory

The literature has related populist storytelling to conspiracy theories (Engesser et al. 2017; Nordensvard and Ketola 2022). The main characteristic of this story is that the populist leader presents himself as the hero, who defends “the people” from various “enemies” who conspire against them. The identity of the enemy changes according to which populist leader is telling the story; for former US president Donald Trump, for instance, the enemy is the establishment in Washington, while Hungary’s prime minister Viktor Orbán singles out the EU bureaucracy in Brussels. As we shall see, AMLO presents himself as a savior and as a hero in the face of diverse conspiracies enacted by the enemies of the people.

In this story, AMLO tells us about a network of actors and institutions that, in a conspiratorial way, oppose the social reforms that his government promotes. According to AMLO, political parties are part of this conspiracy. Two tweets will suffice to convey the plot of this story. First, AMLO links the opposition parties—which, as mentioned above, he pejoratively calls “PRIAN”—with the conspiracy of a former president (Carlos Salinas de Gortari) and its associated corruption: “The bipartisanship of the PRIAN established during the Salinismo is coming to an end. It has only left anti-democracy, corruption, and violence” (Está llegando a su fin el bipartidismo del PRIAN instaurado durante el salinismo y que solo ha dejado antidemocracia, corrupción y violencia: https://t.co/nTkX3T6gBN). The allusions to the PAN and the PRI as allegedly homogeneously corrupt parties and their links with the ancien régime show the nature of this story as a conspiracy theory.

But the opposition political parties are not the only conspirators, according to AMLO. The co-occurrences among the word conservatives, reform (in the context of opposition to), and social (as in social change) (Figure 1, purple) portray the idea of a conspiracy in which conservatives (encompassing political parties, intellectuals, and the press) unite to oppose to the social reforms promoted by López Obrador. Qualitative analysis of AMLO’s tweets shed light on the semantic content of the quantitative co-occurrence between words. Alluding to the supposedly conspiratorial behavior of both politicians and intellectuals, AMLO posts: “Conservatives paid intellectuals to applaud the oppressive regime of corruption, unfairness, and privileges” (Los conservadores pagaban a los intelectuales para aplaudir al régimen opresor de corrupción, de injusticias y de privilegios: https://t.co/KVAgM7n1U5).

However, AMLO goes further: democratic institutions themselves are also cast as conspiring against the people. In a third tweet, according to AMLO, “[t]he verdict of the TEPJF [Electoral Court of the Judicial Power of the Federation] in favor of the ‘independent’ candidate has a lot of substance, but the best thing about the farce is seeing members of the mafia of power, slaves, and spokesmen, tearing their clothes. The true doctrine of the conservatives is hypocrisy” (El fallo del TEPJF a favor del candidato “independiente” tiene mucha miga, pero lo mejor de la farsa es ver a integrantes de la mafia del poder, achichincles y voceros, rasgándose las vestiduras. La verdadera doctrina de los conservadores es la hipocresía: https://t.co/FP6LPnWN3a).

Thus, according to this story, it is not only the opposition political parties, the press, and the intellectual class but also the very checks and balances of representative democracy that are acting in a coordinated manner to put a stop to López Obrador’s attempts to carry out social reform. Accordingly, actors and institutions of all kinds who are different from AMLO and his MORENA party are depicted as partners in a conspiracy against the people.

Invoking the “Heartland”

One element crucial for the understanding of populism as storytelling is its story on “the glorification of the heartland,” that is, an “idealized conception of the community” (Engesser et al. 2017, 1111; see also Neumann-Ernst 2019, 100; Taggart 2000, 274) or a “retrospective utopia” (Priester 2012, 2; Leidig 2019, 118). Thus, “the past plays a paramount role” in this story: “performing the epic function, the distant past”—as opposed to the recent past, when a crisis began—invokes the idea of an original “heartland,” an imagined past in which a “morally impeccable, unified population resides” (Taş 2020, 131; see also Taggart 2000, 274). Typically, populist leaders propose to restore that glorious, lost distant past. The idea of a heartland “is not directed at the future but at the past,” a past that “is not based on rational thoughts or historical facts but deeply rooted on emotions” (Taggart 2000, 95). Hence, “[p]opulists invoke the image of a virtual location which is occupied by the people that represents the ‘core of the community’” (Engesser et al. 2017, 1112–1113; see also Taggart 2000, 96). The ideas of a “Middle America” or “La France Profonde” are emblematic instances of the idea of heartland (Priester 2012; Neumann-Ernst 2019, 100; Taggart 2000, 97).

AMLO’s narrative forms a story that alludes to an idealized community that evokes the idea of a “heartland” common among other populist leaders. López Obrador’s heartland seems to be represented by the indigenous communities. In this vein, our findings (Figure 3) show a strong co-occurrence among the people, indigenous people, small communities, and Oaxaca (light blue). ⁷ In other words, in the tweets in which AMLO uses the word “the people,” he also systematically mentions those terms.

Qualitative textual analysis of AMLO’s tweets shows that his idea of “the people” seems to be synthesized in indigenous communities: López Obrador’s references to indigenous communities are always very positive, and typically suggest that the “true Mexico” is made up of indigenous peoples. One tweet is particularly illustrative of this story: “I met with Claudia Morales, a Wixárika from Jalisco; Olga Santillán, a Tepehuana from the south of Durango and Mónica González, from the Cucapá people of Baja California. They have been proposed for the presidency of CONAPRED [National Council to Prevent Discrimination]. They are authentic representatives of pride, greatness and deep Mexico” (Me reuní con Claudia Morales, wixárika de Jalisco; Olga Santillán, tepehuana del sur de Durango y Mónica González, del pueblo Cucapá de Baja California. Están propuestas para la presidencia del CONAPRED. Son auténticas representantes del orgullo, la grandeza y el México profundo: https://t.co/h5ULRSfmaz).

Executive Aggrandizement

Within AMLO’s storytelling, we identify another story which we call “executive aggrandizement”—that is, a story in favor of a political process wherein the sitting president gradually dismantles institutional checks on his power (Bermeo 2016, 10). This story argues that the expansion of executive power is both compatible with democratic values and necessary for expressing the will of “the people.” Populist leaders like AMLO “strategically use populist speech to present themselves as the champion of ‘the people’, while presenting institutional opposition as a barrier to the popular will” (Bessen 2021, 2). One of the findings of this article shows a network of nodes (Figure 3, green) that systematically links the word “people” with both the word “president,” and the word “government” (i.e., “executive branch” in the Mexican political system).

In this story, AMLO equates the institutions of checks and balances characteristic of democracy with fecklessness and corruption. For this reason, according to the plot of this story, there must be an executive aggrandizement (i.e., an expansion of presidential power) that controls the other branches of governance. For instance, AMLO links autonomous electoral institutions such as the INE (National Electoral Institute) with fraud; according to him, this is why only he himself, and not those “unreliable” institutions, can be the true guarantor of clean elections: “The INE and the TRIFE [Electoral Court of the Judicial Power of the Federation] do not inspire trust. For this reason, we are preparing at 100% for the defense of the vote. After this election there will be, it is our commitment, an authentic democracy and there will no longer be electoral fraud in any of its forms” (El INE y el TRIFE no inspiran confianza. Por eso, nos estamos preparando al 100 para la defensa del voto. Después de esta elección habrá, es nuestro compromiso, una auténtica democracia y ya no habrá fraude electoral en ninguna de sus modalidades: https://t.co/eOVsUranDC).

Personal Action Frames

Populist messages “are frequently more personal and sensationalistic in nature,” a characteristic of populist storytelling that has been called the “personal action frame” (Neumann-Ernst 2019, 101; see also Engesser et al. 2017, 1113). In addition, social media have facilitated the use of such personal action frames (Bennett and Segerberg 2012, 744), which has allowed populist leaders to employ more and more personalistic and action-based messages appealing to the use of words or sentences such as the people, I, we, my government, I will finish with, and so on. We found that AMLO’s storytelling on social media, “manifest[s] itself in the shape of a personal action frame” (Engesser et al. 2017, 1114; see also Neumann-Ernst 2019, 101). In this vein, findings show a strong co-occurrence between, on the one hand, we are going to, put an end to, to act, zero impunity, and corruption, and, on the other hand, between uproot, root, and privileges (Figure 3, orange lines).

Qualitative textual analysis confirms the presence of this personal action frame in AMLO’s storytelling. Hence, López Obrador posts that “The Fourth Reform will uproot the corrupt regime of injustices and privileges” (La Cuarta Transformación arrancará de raíz al regimen corrupto de injusticias y privilegios: https://t.co/zITA4wgaUJ). The objective that this story that we call “personal action frame” frequently poses is to end corruption and privileges.

AMLO’s Stories and Polarized Replies on Social Media

Having identified the stories that AMLO uses in his tweets, we proceed to explore their respective relationships with the level of negative affective polarization in the replies that followers upload to his tweets. We measure the level of affective polarization for each story as follows: each story has a percentage of polarization in the replies as an average of the proportion of polarized responses to each tweet in each story. Figure 4 shows the distribution of the percentage of polarized replies to each story, expressed as a violin plot.

Figure 4. AMLO Stories and Negative Affective Polarization [n = 59,561(replies)]

The bodies of the violins suggest that the proportion of polarized replies varies widely according to the type of story. In the story of conspiracy (i.e., “Conspiracy theory”), a good number of the tweets have a very high percentage of polarized replies, exceeding 75% of all replies in many cases (i.e., above the 75 percentile). This suggests that this story tends to result in greater affective polarization. In a different way, the replies to stories in the categories “Invoking the ‘heartland’” and “Personal action frame” generally demonstrate a lower percentage (below 25%, in many cases) of polarized replies, suggesting that these stories tend to produce less affective polarization.

Regarding the measurement of polarization, we also employ a method that calculates the level of affective polarization as a simple count of the replies by each tweet. Figure 5 shows the data on the total number of replies compared with the number of polarized replies by each of AMLO’s stories. Although the findings are not definitive (chi-square = 72; p = 0.13), it is possible to infer that some stories produce a greater number of replies than others and that such stories are those that tend to have a greater number of polarized replies.

Figure 5. Number of Polarized Replies and Number of Total Replies

Additional analysis from descriptive statistics seems to confirm that different stories produce different levels of affective polarization. The range of averages observed in the percent of the polarization variable is 9.21, with the highest value being the story of “Conspiracy theory” (60.165%) and the lowest value being the story of “Invoking the ‘heartland’” (50.954%). See Figure 6, above.

Figure 6. AMLO’s Stories and Levels of Negative Affective Polarization

In order to identify whether there are statistically significant differences between the means of the levels of affective polarization when comparing different story types, we applied the analysis of variance (ANOVA) technique (Table 2). In contrast to the F test, the ANOVA test allows the evaluation of multiple groups and the crossover between all possible pairs of means of the set of observed populations. Thus, the ANOVA test uses the study of variances to identify the presence or absence of a difference in means in at least one pair of observed means.

Table 2. ANOVA Test

The p-value of the ANOVA test does not constitute definitive evidence allowing us to reject the Ho of difference between means based on the variances for the means of the groups studied (p-value = 0.082). However, this result may be a consequence of the large differences in the size of the groups, given that the descriptive evidence (i.e., the violin plot in Figure 4) suggests that different stories produce differences in levels of negative affective polarization.

To visualize the comparisons between the different pairs of means that were compared in the ANOVA, we applied a Tukey test (Table 3). This test allows us to see the results for each pair and makes analytical sense if there is enough evidence to reject H0 and observe in which pairs the statistically significant differences are found. In the obtained results, there is an average difference of 6.12, indicating that, on average, the differences between the means of the compared groups are relatively small. However, this value should be interpreted with caution due to the presence of extreme values. Regarding the standard deviation, there is significant variability in the mean differences between the comparisons. This suggests that some comparisons show much larger differences than others.

Many of the adjusted p-values are high, indicating that, under Tukey’s adjustment, we did not find statistically significant differences between the stories. Although there may be some comparisons of means that appear to have a significant difference, when we consider their standard deviation—which is part of what the test considers to determine significance —this is not the case due to the wide dispersion.

Table 3. Tukey Test to Visualize the Pairs of Compared Means

To deepen the portrait that quantitative evidence provides, we finally explore the types of affective polarization that each of these stories produces—that is, the differences or similarities among the words used in the polarized replies as they correspond to each of the story types that AMLO uses in his tweets. This exercise also seeks to identify whether the polarized replies come primarily from supporters or opponents of AMLO.

Figure 7 shows our results. On the right-hand side of this figure, you can see the gray node indicating the frequency of the words (F). The orange nodes represent the eight story types of AMLO; the blue nodes correspond to the words in the replies that are repeated in front of tweets that belong to different stories; and the purple nodes symbolize the words in the replies that are specific to a single story (i.e., they are not repeated in other stories). Thus, this analysis seeks to illuminate the qualitative differences between the replies according to the different story types—that is, to determine whether the different story types produce differences or similitudes in the content of negative affective polarization.

Figure 7. AMLO’s Stories and Types of Negative Affective Polarization

Findings show that there are polarized words in the replies shared by different AMLO stories (Figure 7). This occurs, for instance, with the words corrupt, thief, shit, and motherfucker, words that followers post in front of tweets that belong to different AMLO stories. There are also, however, other polarized words that are only used in front of a specific story type and are not found in others: for example mafia (conspiracy theory), bitch (ostracizing the others), ridiculous (democracy as direct democracy), evil (our translation of demonio, literally “demonic”: corrupt and exploitative elites), and quack (our translation of charlatán: executive aggrandizement).

A suggestive finding is that the term “corrupt” has the highest frequency among the replies of the followers of AMLO’s account; indeed, it appears associated with most of the story types that López Obrador uses. In other words, the replies to the different stories have in common the use of the term “corrupt.” Qualitative textual analysis of the polarized replies indicates that this word is used mainly by pro-AMLO Twitter followers, suggesting that AMLO’s storytelling successfully persuades his supporters that his opponents are all homogeneously corrupt.

Conversely, the word PRIAN is present only in the replies to AMLO’s tweets that belong to two story types that, interestingly, engender the highest polarization: the story of conspiracy (60.165%) and the story of ostracizing (56.586 %). Furthermore, the word mafia appears only in the story type with the highest level of negative affective polarization, the story of conspiracy. Qualitative textual analysis suggests that this story of conspiracy seems to have a great influence in generating the idea of seeing the opposition parties (i.e., “PRIAN,” in AMLO’s coinage) as part of a conspiring “mafia” and, therefore, accentuating among the supporters of AMLO the idea of the “us versus them” view of the political world that is typical of negative affective polarization.

Final Remarks

The narrative dimension of populism has largely been ignored (Aalberg et al. 2016; Arzheimer and Carter 2006; Rydgren 2005). Conceding that “populist leaders are persuasive storytellers,” this article aims to discover the overlooked “dynamics of populist meaning-making” (Taş 2020, 128) for the case of AMLO in Mexico. The findings of this work contribute to emphasizing the importance of storytelling in understanding the phenomenon of populism and its consequences—such as negative affective polarization. The consequences of populism have become the subject of new and developing literature that has recently explored the links between populist narrative and affective polarization (Wiesehomeier et al. 2024; Sarsfield 2024). As a background to these recent studies, some scholars have already suggested the idea of intentional or deliberate polarization by populist leaders (e.g., Corrales 2005; Szymański and Cihangiroğlu 2023), a hypothesis that still requires further research. While this study is exploratory and theory-generating, its results contribute to a better comprehension of the logic of populism, providing evidence that supports the expectation that populism leads to affective polarization.

In this vein, the findings of this work reinforce existing research in favor of the idea that certain narratives among elites play an important role in driving negative affective polarization among citizens (e.g., Druckman et al., 2021; Gentzkow et al. 2019; Lau et al. 2017). At the same time, the study finds evidence in favor of the idea that the different AMLO stories seem to induce different types of negative affective polarization in the replies on Twitter. For instance, “corrupt” is quite a common reply to all AMLO’s populist stories in Twitter, while “PRIAN” as a reply only appears in two of his stories, the “conspiracy theory” and “ostracism” stories. These findings suggest that the type of polarized reply depends on the story told by the AMLO posts on Twitter. These results generate new hypotheses to be tested in the future.

This article thus tries to explore what, following a growing literature on political rhetoric, we call “storytelling” (e.g., Engesser 2017; Nordensvard and Ketola 2022; Seargeant 2020; Ungureanu and Popartan 2020). We simply define storytelling as the “art of telling a story where emotions, characters and other details are applied” in order to “promote a particular point of view or set of values” (Nordensvard and Ketola 2022, 861). Focusing on an exploratory research study, this work seeks to define the kinds of stories that AMLO uses in his Twitter account and identify which of these stories provokes the greatest polarization among his followers (both his supporters and his opponents) on social media. The different stories that López Obrador uses seem to have effects on both the levels and the types of affective polarization. On the one hand, findings suggest that the story types we call “conspiracy theory” and “ostracizing the others” push individuals to higher degrees of negative affective polarization. On the other hand, the content of polarized replies vary with the different kinds of stories that AMLO uses in his tweets.

In other words, the findings of this study suggest that the specific stories spread by AMLO on social media contribute to an increased negative affective polarization among the followers of his account and give rise to various dimensions of negative affective polarization. However, these results pose a new research challenge of examining whether this phenomenon is also observed with other populist leaders or if non-populist leaders also generate negative affective polarization. These are questions that remain open for future research.

Our article finds that certain narratives among elites play a significant role in driving negative affective polarization among the masses (Druckman et al., 2021).⁸ Additionally, our results suggest that affective attitudes in the public can be directly influenced by the tone of political elites (Gentzkow et al. 2019) and that diverse media environments and negative campaign rhetoric can result in higher levels of negative affective polarization (Lau et al. 2017). In particular, we focus on how what we call “populist storytelling” presented by these elites on social media contributes to this phenomenon. Our results seem to coincide with these previous works, highlighting the influence of both the tone of political elites and negative campaign rhetoric on negative affective polarization.

At this point in the work, an important caveat is acknowledging its limitations. Regarding the generalizability of the findings, it is important to emphasize that a limitation of this study is that it is based on data from the Mexican case. The answer to the question of how generalizable the results of this case are will depend on further research conducted on other cases. The contribution to our understanding of the relationship between populism and negative affective polarization is limited to—but also consists of—providing some findings that support some of the hypotheses explored in the literature for other cases (e.g., Druckman et al., 2021; Lau et al., 2017) as well as proposing new hypotheses to be tested in future research.

A second limitation of our study is that we do not have reliable and extensive access to the demographic characteristics of AMLO’s account followers. Therefore, we have not been able to test whether the type of followers responding to different stories is similar. Additionally, we have not been able to determine whether the timing of AMLO’s stories is somehow related to more or less polarized contexts. We acknowledge that alternative measures and robustness checks are necessary for further research.

A third limitation of our work emerges from the representativeness of the people who use Twitter and the validity of our results compared to those we would obtain if we had used other social media. While Twitter provides a rich environment for exploring the nexus between populist storytelling and negative affective polarization, it is vital to recognize the potential limitations in terms of the representativeness of our findings. We are aware of these limitations given the distinct demographic composition of its users, which is not representative of the population. At the same time, and due to feasibility reasons, we were not able to explore other social media. We only study AMLO’s stories and the replies of his followers within Twitter, with no discussion of whether our conclusions could depend on the media. However, we believe that the selection of Twitter as our primary social media platform is justified if we consider that our research question aligns with that platform’s inherent political nature. Twitter has evolved into a prominent political arena characterized by unique features such as character limits, rapid information dissemination, and direct engagement of political leaders with their followers. These characteristics have been recognized by scholars (e.g., Aalberg et al. 2016; Lau et al. 2017; Urbinati 2019), emphasizing that Twitter facilitates political information diffusion.

Finally, our results speak in favor of the need to expand the research agenda on the effects of storytelling by populist leaders on the attitudes of citizens, particularly focusing on negative affective polarization. Due to the exploratory nature of this work, further research is required, particularly through studies that employ an experimental design. This will allow testing the findings based on the observational data from this study to determine whether the co-occurrence between AMLO’s stories and level of negative affective polarization is indeed causal. The significance of this work lies in its ability to provide findings that contribute to the advancement of experimental designs of this kind. To the extent to which negative affective polarization hinders cooperation among individuals and inhibits the construction of the kinds of agreements typical of democracies, it is important to study the effects of certain stories put forth by political leaders. Otherwise, there is a risk of deepening the erosion of democratic institutions and values, a phenomenon which, a broad consensus agrees, is growing in contemporary democracies.