INTRODUCTION
“I did not get history rewritten. I just got it rectified.”
–Murli Manohar Joshi, as quoted in Pathak (Reference Pathak2019)
Shortly following Hindu nationalist Bharatiya Janata Party’s rise to power in India in 1998, observers noted an elemental and sweeping shift in the state’s educational content: secular narratives of the nation’s history that highlighted its pluralism had in many places been replaced with accounts that portrayed India as a fundamentally Hindu nation to which Muslims had arrived as a “foreign” and “invading” force. Critics alleged that the changes were part of a radical effort by Education Minister Murli Manohar Joshi and his hand-picked allies at the National Council for Education Research and Training to rewrite Indian history to fall in line with Hindutva, a nationalist ideology to which Joshi subscribed grounded in notions of Hindu supremacy (Thobani Reference Thobani2019). In the series of “education wars” that followed, Joshi and his defenders in turn responded that they were merely correcting the historical account.
As in India, activists and leaders across many countries today are locked in battles over which version of history should be promoted by the state and thereby which groups should be remembered, and how, in collective memory. Particularly visible are debates in the United States and in Europe over statues, street names, and public symbols that honor leaders and movements that committed atrocities against marginalized communities (Tharoor Reference Tharoor2020). These politics of the past suggest that there are different ways a country can portray its history and that political actors have a stake in presenting a particular narrative (Nussbaum Reference Nussbaum2009; Reicher and Hopkins Reference Reicher and Hopkins2001). Yet even as battles intensify, mobilizing passionate supporters on both sides, we still know little about the political effects of different historical representations.
Do different representations of a nation’s history affect which social groups seek out, and ultimately obtain, positions of power? In particular, can historical narratives that exclude marginalized communities deepen their underrepresentation in leadership roles? And could highlighting marginalized group members’ historical contributions to the nation encourage their political inclusion? We provide a theoretical framework and an empirical test aimed at increasing extant understandings of the link between historical representations and the marginalization of historically disadvantaged groups, such as ethnic minorities.Footnote 1 We hypothesize that exclusionary historical representations can undercut marginalized communities’ perceived claim to the nation and entitlement to speak on behalf of its “unquestioned” members, and thus make marginalized individuals less willing to seek out leadership positions (supply channel) and greater society less disposed to support them when they do (demand channel). Consistently, we posit that an inclusive representation can serve to increase marginalized group members’ perceived belonging and entitlement to make decisions that affect others in the nation. We expect our theory to potentially apply to minority representation in all positions that require leaders to speak on behalf of social groups whose belonging to the nation is “beyond question” and not only political offices.
Our theory is rooted in the notion that history serves a crucial function in defining and legitimizing boundaries of national belonging and can therefore be used to promote politics of inclusion or exclusion (Liu and Hilton Reference Liu and Hilton2005; Sibley et al. Reference Sibley, Liu, Duckitt and Khan2008). According to Wodak (Reference Wodak2009, 25), “Historical memory […] is an indispensable prerequisite for national identity […] The further into the past, the real or imaginary memories reach, the more securely national identity is supported.” Tying together insights from literatures on history and the nation with work on entitlement and leadership emergence, we posit that history has both a descriptive and a prescriptive dimension: by describing who has participated in and contributed to the nation historically, prescriptions regarding who is entitled to make decisions on its behalf can be derived. Historical narratives that depict a group as having long-standing ties to a nation and positively contributing to its development might cast that group as part of the nation’s core constituency endowed with a mandate to lead. In contrast, individuals from groups depicted as negative and “foreign” entities could be classified as peripheral with a mandate to follow.
We test our theory through a pre-registered, incentivized online experiment with 1,592 respondents in India, where Muslims are heavily underrepresented across a wide range of important civil society, economic, and political leadership positions and there is an active movement to minimize Muslim historical contributions and recast the country as a Hindu nation. In our experiment, Hindu and Muslim participants are tasked with completing several exercises sourced from real, official Indian school textbook material, for which they can earn additional incentives. We randomly assign participants a history exercise that either highlights the plural nature of Indian history and Muslims’ historical contributions to the nation (inclusive representation treatment, promoted by the secular left), describes a glorious Hindu past and a dark period of Muslim invasions (exclusive representation treatment, endorsed by the Hindu right), or which details unrelated history on topics such as agriculture and the railway system (baseline condition, not advocated for by one more than the other).
Following the history exercise, participants are randomly matched in groups of three and are told they must complete two additional school exercises with payoffs determined by group performance and all members earning an equal share. While they will complete both additional exercises, only one will count for payment for all group members, in accordance with the decision of a group representative. Our central outcomes of interest from this set of decisions are a participant’s (1) willingness to volunteer as group representative (supply channel, elicited before participants became aware that the game had a demand component) and (2) their preference ranking of their group partners as group representative (demand channel). Our rich set of outcome data allow us to evaluate our theory using both stated and revealed preferences, and using both an incentivized outcome over which we have a high degree of experimenter control as well as outcomes over which we have less control but which offer more clear real-world corollaries.
To evaluate our hypotheses, we consider whether, and how, views about individuals from a marginalized group (Muslims) and their suitability for leadership roles, as well as their willingness to seek out and of society to grant them such roles, change depending on respondents’ random assignment to a historical representation (inclusive, exclusive, or baseline). Our findings reveal that historical representations can affect the perceived centrality of different groups to the nation: while exclusive renderings of history increase the perceived centrality of Hindus to the Indian nation, inclusive renderings increase the perceived centrality of Muslims. We also observe that historical representations can affect the supply of and demand for ethnic minorities in leadership positions. Muslim respondents who are exposed to inclusive representations of national history are more eager to take on a leadership role than are Muslim respondents exposed to an exclusive or a more neutral version of history. Inclusive representations also positively affect Muslim respondents’ perceptions about their own and real-world Muslim politicians’ entitlement to lead, and their demand for real-world Muslim politicians. While these findings highlight the potential of inclusive narratives to increase minorities’ leadership aspirations for themselves and members of their community, Hindus are largely unmoved by the inclusive treatment in their demand for Muslim leaders. We conclude that shifting majority groups’ views on minority leadership may be particularly challenging.
To the best of our knowledge, our study represents the first quantitative assessment of the link between historical narratives and the supply of and demand for historically disadvantaged groups in leadership positions. As such, our paper sheds new light on the tangible consequences of symbolic politics, a topic of interest far beyond the Indian context with implications for ongoing real-world debates. Previous and ongoing work examines the effects of changes to symbols and narratives of nationhood—conveyed via mediums including textbooks, statues, street names, and flags—on national identity (Chen, Lin, and Yang Reference Chen, Lin and Yang2023; Durrani and Dunne Reference Durrani and Dunne2010), political attitudes and beliefs (Balcells, Palanza, and Voytas Reference Balcells, Palanza and Voytas2022; Cantoni et al. Reference Cantoni, Chen, Yang, Yuchtman and Jane Zhang2017; Rahnama Reference Rahnama2021), criminal activity (Rahnama Reference Rahnama2021), electoral outcomes (Rozenas and Vlasenko Reference Rozenas and Vlasenko2022; Villamil and Balcells Reference Villamil and Balcells2021), and even housing prices (Green et al. Reference Green, Jame, Lee and Lee2022). We contribute to this growing literature by exploring whether different state-sponsored historical narratives shape marginalized group members’ leadership aspirations, and greater society’s acceptance of their leadership. In a time of rising ethnonationalism across the globe, where majorities claim fusion with the nation and use the power of the state to exclude minorities from symbolic representations of nationhood, this question could not be more pressing.
Our research provides both theoretical and methodological contributions. By tying together heretofore distinct lines of study on social representations of history (Liu and Hilton Reference Liu and Hilton2005; Reicher and Hopkins Reference Reicher and Hopkins2001; Sibley et al. Reference Sibley, Liu, Duckitt and Khan2008), entitlement (Liddle and Michielsens Reference Liddle, Michielsens and Rai2000; O’Brien and Major Reference O’Brien, Major, Jost, Kay and Thorisdottir2009), and leader emergence (Conroy and Green Reference Conroy and Green2020; Norris and Lovenduski Reference Norris and Lovenduski1995), our theory offers new insights into the mechanisms through which historical representations can shape the supply of—and demand for—marginalized groups in leadership roles. Our experimental design employs innovative methods to explore these mechanisms empirically, through the use of incentivized exercises and group tasks among participants represented by religiously identifiable avatars. In doing so, we follow recent calls to make surveys more engaging for participants by making them more “game-like” (Salganik Reference Salganik2019).
Finally, our paper contributes to additional strands of literature rarely in conversation. First, our research speaks to work on the consequences of different founding narratives of nationhood (Mylonas and Kuo Reference Mylonas, Kuo and Thies2017; Singh and Vom Hau Reference Singh and Vom Hau2016; Straus Reference Straus2015), in particular, as conveyed through school content. For instance, King (Reference King2013) presents qualitative evidence indicating that school content—especially history curricula—can produce and amplify exclusive identities, potentially enhancing social divisions and promoting violence. We build on this literature by examining the effects of exclusive and inclusive historical narratives, conveyed through textbook material, on minorities’ leadership aspirations and greater society’s acceptance of minority leadership. Second, we add to work examining exclusion, national belonging, and political behavior among ethnic minorities. Our findings offer new and important insights into how minorities respond, socio-politically, to dynamics of exclusion (Abdelgadir and Fouka Reference Abdelgadir and Fouka2020; Simonsen Reference Simonsen2021), under what conditions exclusion mobilizes or demobilizes (Hobbs and Lajevardi Reference Hobbs and Lajevardi2019; Oskooii Reference Oskooii2020; Weiss, Siegel, and Romney, Reference Weiss, Siegel and Romney2022), and how minority inclusion can be encouraged (Alrababa’h et al. Reference Alrababa’h Ala’, Mousa and Siegel2021; Williamson et al. Reference Williamson, Adida, Lo, Platas, Prather and Werfel2021). Third, we extend research on the underrepresentation of historically disadvantaged groups by theorizing a new factor that might contribute to the sustaining (or reversal) of marginalization: historical representations (Bueno and Dunning Reference Bueno and Dunning2017; Dancygier et al. Reference Dancygier, Lindgren, Nyman and Vernby2019; Portmann and Stojanović Reference Portmann and Stojanović2019). We test for both supply and demand channels driving underrepresentation, an advance on a body of research that has primarily focused on demand-side explanations. We speak to a growing literature exploring how supply-side channels such as socialization (Bos et al. Reference Bos, Greenlee, Holman, Oxley and Celeste Jay2022), ambition (Bonneau and Kanthak Reference Bonneau and Kanthak2020; Fox and Lawless Reference Fox and Lawless2014), confidence (Wolak Reference Wolak2020), and election aversion (Kanthak and Woon Reference Kanthak and Woon2015) might explain marginalized group underrepresentation.
THEORY AND HYPOTHESES
Do different portrayals of a nation’s history affect whether members of historically marginalized groups seek out, and ultimately obtain, positions of power? A large literature has identified a number of demand- and supply-side factors that influence, both independently and in interaction with each other, who becomes a leader (Gulzar Reference Gulzar2021; Holman and Schneider Reference Holman and Schneider2018). This literature has produced valuable insights but further questions remain, in particular, as the underrepresentation of marginalized groups in leadership positions persists despite changes in theorized relevant factors such as the prevalence of overt discrimination, political party and organizational gatekeepers, institutional barriers, and the composition of the electorate (Gulzar Reference Gulzar2021; Lawless Reference Lawless2015).
In this paper, we propose an additional factor that we hypothesize may contribute to representation outcomes through both demand- and supply-side channels: historical representations. Specifically, we argue that historical representations—namely, if and how different groups are represented in collective memory—can alter which individuals are perceived as belonging to a nation’s core constituency, entitled to represent the interests of other members of the nation. As a consequence, we posit, historical representations can have important implications for who is willing to seek out leadership positions, as well as who is more likely to be accepted and supported as a leader by greater society. Taken together, our theoretical framework suggests that different depictions of history can serve as one contributing factor affecting the level of descriptive representation of various social groups in positions of power.
In this section, we present our theory and pre-registered hypotheses.Footnote 2 First, we detail how historical representations can affect understandings of different groups’ place in the nation. Second, we explain how the degree to which marginalized groups are viewed as “peripheral” to the nation can carry consequences for whether their members are perceived as entitled to speak on behalf of the nation’s “core” constituency. Finally, we argue that these developments can affect whether marginalized individuals ultimately adopt leadership positions requiring them to speak on behalf of “core” members of the nation through two channels: first, through affecting these individuals’ desire to lead (supply channel), and second, by influencing greater society’s willingness to support them as leaders (demand channel).
Historical Representations, the Nation, and Who Can Claim It
The first component of our theory asserts that historical representations can play a critical role in defining current conceptions of national identity and different groups’ perceived claims to the nation, especially when reinforced or promoted by entities exerting authority over the nation (e.g., the state and its representatives). History is recognized by many scholars as a key ingredient in the creation and maintenance of the imagined community of nationhood, as well as for other socially constructed identity categories (Anderson Reference Anderson1983; Hobsbawn Reference Hobsbawn1990). While nations may differ in the degree to which formal membership is explicitly tied to the past (notably, some countries ascribe citizenship by descent, others by civic criteria; see Brubaker Reference Brubaker1992; Chandra Reference Chandra2012), most if not all boast an origin story and historical narrative that helps to define a national identity and distinguish “us” from “them” (Brubaker et al. Reference Brubaker, Feischmidt, Fox and Grancea2006; Horowitz Reference Horowitz1985; Straus Reference Straus2015). Indeed, initial claims for nationhood were often justified by the assertion of “historic rights” (Brubaker et al. Reference Brubaker, Feischmidt, Fox and Grancea2006) or a shared “mythic history” (Laitin Reference Laitin1995).Footnote 3
Building on this literature, we understand historical representations as one way that people draw boundaries between the nation’s “core” (the essential, natural, and unquestioned members) and those relegated to the “periphery”; those “who might be allowed to be part of the nation, but ‘never quite’” (Pandey Reference Pandey1999, 608).Footnote 4 We identify two main pathways through which history can affect different groups’ perceived “centrality” to the nation, and thus, their right to claim membership in its core community.Footnote 5
First, through the incorporation of certain groups (and not others) into stories of national identity, historical narratives can define who is a “prototypical” in-group member and who is not (Reicher and Hopkins Reference Reicher and Hopkins2001; Sibley et al. Reference Sibley, Liu, Duckitt and Khan2008). Narratives can further delineate these groups by establishing “foundational myths” that endow prototypical in-group members with a shared ancestry, common traits, or a “golden age” (Guichard Reference Guichard2010). Second, by highlighting or downplaying different groups’ contributions to the nation, historical narratives can legitimize or delegitimize a society’s current or projected social and political arrangements, including which groups are thought to have a “right” to core membership (Khan et al. Reference Khan, Svensson, Jogdand and Liu2017; Straus Reference Straus2015). For instance, narratives that undermine a group’s historical contributions, omit them entirely, or depict them as “enemies of the nation,” can justify that group’s exclusion from the national community (Khan et al. Reference Khan, Svensson, Jogdand and Liu2017). Faced with such narratives, individuals might internalize and accept the view that presently dominant groups are core members of the nation, and that marginalized communities are peripheral. Historical narratives can thus act as “legitimizing myths,” providing intellectual and moral justifications for unequal social practices and helping to rationalize power differences between groups in society (Jost, Banaji, and Nosek Reference Jost, Banaji and Nosek2004; Sidanius and Pratto Reference Sidanius and Pratto1999). As suggested by Social Dominance Theory (Jost, Banaji, and Nosek Reference Jost, Banaji and Nosek2004), disadvantaged group members often search for precisely such justifications of the existing social order and, upon finding them, internalize a sense of inferiority.Footnote 6
In sum, we posit that narratives can affect perceptions of whether individuals from marginalized groups “truly” belong to the nation through either of two main pathways: one, by affecting whether they are thought to fit the criteria for membership, and the second, by affecting whether they are believed to have “earned” the right to be part of its core community. An inclusive historical representation that makes space for a social group in the nation and positively highlights its contributions establishes that group’s centrality to the nation and the belonging of its individuals as core, or full, members. In contrast, an exclusive historical retelling that omits or actively attributes a group a negative role in the nation’s history suggests that individuals from that group either do not belong in the nation or do not belong fully (i.e., are situated outside the national identity boundary or inside but on the “periphery”).
$ {H}_1 $
: Historical representations can shape the perceived prototypicality and historical contributions (“centrality”) of different social groups to the nation. Specifically:
-
(a) Exclusive historical representations will reduce the perceived centrality of marginalized groups to the nation.
-
(b) Inclusive historical representations will increase the perceived centrality of marginalized groups to the nation.
Claim to the Nation and Entitlement
The foregoing discussion brings us to our central theoretical claim: we posit that through shaping perceptions of who is a prototypical member of the nation and who has contributed to the nation, historical representations can alter perceptions of who is entitled to speak on behalf of the nation’s “core” constituency. We define “entitlement” as the notion that a person, or category of people, should enjoy a particular set of rights by virtue of who they are (O’Brien and Major Reference O’Brien, Major, Jost, Kay and Thorisdottir2009). One can feel entitled to a nation’s material resources such as welfare or jobs and also to less tangible psycho-social benefits, such as the right to judge others’ behavior or make decisions that will affect other members of the nation (Skey Reference Skey, Jackson and Molokotos-Liederman2014).
Qualitative research provides suggestive support for our posited tie between a group’s perceived centrality to the nation (their prototypicality and recognized historical contributions) and whether its members are viewed, and view themselves, as entitled to speak on the nation’s behalf. Reicher and Hopkins (Reference Reicher and Hopkins2001) posit that the more an individual conforms to a prototypical group member in terms of traits, norms, and behavior, the greater their perceived right to speak in the group’s name. Debates on immigration routinely feature arguments that certain citizens “deserve” more than others due to their prior contributions to the nation (Sainsbury Reference Sainsbury2012) and ethnic minorities themselves have been found to link their status within the nation with their willingness to voice critical reflections on society (Hopkins and Blackwood Reference Hopkins and Blackwood2011; King Reference King2013; Simonsen Reference Simonsen2022). In particular, disadvantaged groups who view national hierarchies as legitimate often come to exhibit a depressed sense of entitlement relative to members of advantaged groups (Jost, Banaji, and Nosek Reference Jost, Banaji and Nosek2004; Major Reference Major1994). Narratives that call into question a group’s historical contributions could therefore “create group differences in perceived entitlement by enhancing the degree to which differential outcomes are seen as legitimate reflections of differential inputs” (O’Brien and Major Reference O’Brien, Major, Jost, Kay and Thorisdottir2009, 433). In sum, by affecting perceived belonging, historical representations may affect different social groups’ perceived entitlement, “so that ‘I belong more than you’ also means ‘I deserve more than you’” (Skey Reference Skey, Jackson and Molokotos-Liederman2014, 327).
$ {H}_2 $
: Historical representations can shape the perceived entitlement of different social groups to speak on behalf of the nation. Specifically:
-
(a) Exclusive historical representations will reduce the perceived entitlement of marginalized groups to speak on behalf of the nation.
-
(b) Inclusive historical representations will increase the perceived entitlement of marginalized groups to speak on behalf of the nation.
Entitlement and Leadership
The final building block of our theory posits that perceived entitlement to speak for the nation may have consequences for who ultimately does speak for the nation. Specifically, we theorize that entitlement can affect individuals’ willingness to seek out leadership positions (supply mechanism), and of society to grant them the opportunity to lead (demand mechanism). Our theory draws on studies of leader emergence, which indicate that individuals who are perceived or perceive themselves as less entitled to lead can face significant internal and external hurdles when considering or attempting to obtain such leadership roles.
Individuals perceived as less entitled to lead may face questions about their legitimacy and deservingness to serve in leadership roles, thus weakening their incentives to put themselves forward as candidates. Research suggests that aspiring leaders who do not view themselves as “natural leaders” struggle to gain acceptance in representative roles. As Liddle and Michielsens (Reference Liddle, Michielsens and Rai2000, 128) note, if entitlement is not self-evident, “a sense of entitlement has to be both constructed for the self and publicly presented to others before the authority to exercise power is recognized.” Relatedly, individuals who perceive themselves as less entitled often experience doubts about their competence as leaders. Norris and Lovenduski (Reference Norris and Lovenduski1995) observe that a lack of self-confidence in the ability to lead or feelings of being “unqualified” can weaken marginalized group members’ willingness to seek out leadership positions.
Perceptions of entitlement are also expected to shape greater society’s acceptance of marginalized groups in leadership roles. We suggest that dominant groups should be less likely to accept decisions made by marginalized group members when these groups are cast as less central to national identity and thereby less entitled to speak on their behalf. As a result, marginalized group members should also find less support among greater society when pursuing leadership roles, if perceived as less entitled. For instance, beliefs about deservingness often underlie arguments against affirmative action policies aiming to increase the representation of marginalized communities: critics among greater society often contend that empowered individuals would be less competent leaders, implying that there are not enough marginalized community members suited to fill government or other positions (Gulzar, Haas, and Pasquale Reference Gulzar, Haas and Pasquale2020; Jensenius Reference Jensenius2015).
In sum, we expect that historical representations can endow certain groups with a greater claim to the nation, distinguishing “those who are recognised as having a legitimate entitlement to judge who and what is appropriate within the bounded territory of the nation,” from those who are not (Skey Reference Skey, Jackson and Molokotos-Liederman2014, 109). As a result, we propose that exclusive (inclusive) historical representations and resultant effects on belonging and entitlement could contribute to there being fewer (more) disadvantaged group members in leadership roles both through altering marginalized group members’ willingness to lead and by changing greater society’s receptivity to their leadership.Footnote 7
$ {H}_3 $
: Exclusive historical representations will reduce the presence of marginalized group members in leadership roles by:
-
(a) reducing the willingness of marginalized group members to seek out leadership roles (supply channel),
-
(b) reducing the willingness of greater society to grant them such roles (demand channel).
$ {H}_4 $
: Inclusive historical representations will increase the presence of marginalized group members in leadership roles by:
-
(a) increasing the willingness of marginalized group members to seek out leadership roles (supply channel),
-
(b) increasing the willingness of greater society to grant them such roles (demand channel).
According to our theory, the effects of historical representations operate primarily through changing perceptions of who is a “core” member of the nation, and thus who is entitled to take decisions on behalf of such “true” members. Since historical representations are expected to alter perceived hierarchies of belonging, changes in marginalized group members’ willingness to lead should mainly be observed when they are asked to speak on behalf of members of a different perceived “rank” within the nation. Our final hypothesis reflects this expectation.
$ {H}_5 $
(Group Composition): Effects of historical representations on marginalized group members’ willingness to seek out leadership roles will be greater when the leadership position requires that marginalized group members speak on behalf of “unquestioned” core members of the nation, than where the position requires that they speak only on behalf of other marginalized group members.
Summing Up: Historical Narratives and Who Speaks for the Nation
As textbooks are re-written, streets are renamed, and different heroes are celebrated, new imagined boundaries of the nation crystallize. We argue that this can have implications for who seeks out, and who obtains, positions of power. Table 1 summarizes our central hypotheses and illustrates how we expect exclusive and inclusive renderings of history to affect the perceived centrality, entitlement, supply of, and demand for leaders from marginalized communities, among both marginalized and dominant group members.Footnote 8
Summary of Hypotheses
Note: Table presents a summary of our hypotheses and associated empirical predictions. Our theory does not result in fully symmetric empirical expectations for marginalized and dominant groups. As explained in footnote 7, references to “greater society” in demand hypotheses
$ {H}_{3b} $
and
$ {H}_{4b} $
refer to the full population, inclusive of both marginalized and non-marginalized individuals; predicted effects for these hypotheses are thus labeled as relevant for “all” individuals. * refers to the moderating effect predicted by
$ {H}_5 $
, that historical representations will have a larger effect on marginalized individuals’ supply of leadership where a leadership position requires that they speak on behalf of “core” members of the nation.
SETTING: NATIONALISM, REVISIONISM, AND LEADERSHIP IN INDIA
We conduct an experimental test of our theoretical argument in India, where Muslims are both underrepresented in important leadership positions and their role in the nation’s history and the national community are under active contestation.
The Rise of Hindu Nationalism and Historical Revisionism
In India, two radically different, yet fully developed accounts of the nation’s history coexist uneasily. On the one end, many on the secular left (notably, the Indian Congress Party [INC]) have long defended a version of history that emphasizes the plural nature of the Indian nation. This historical narrative has served to promote an inclusive conceptualization of the Indian nation, based on civic values (such as tolerance and respect for democratic principles) rather than ethnic traits, as well as a common history and the shared goal of modernity. This secular (or “Nehruvian”) tradition embraces the heterogeneity of India’s traditions and perceives it as central to Indian identity—constituting a strength rather than a weakness (Nussbaum Reference Nussbaum2009). On the other end, members of the Hindu right (prominently today, the ruling Bharatiya Janata Party [BJP]) promote a version of history in which Hindus have a unique claim to the Indian nation and which excludes and vilifies the country’s Muslim minority population (Thobani Reference Thobani2019). This narrative has served to undergird an exclusionary, “ethnic” understanding of India according to which Hindu and national identity perfectly overlap.Footnote 9
The “Textbook Controversy”
While we expect our theory to apply to historical narratives conveyed through a number of different possible mediums, we elected to study the effects of content taken from school textbooks for a few reasons. First, textbooks can offer a special lens into narratives advocated for by state actors—changes under different educational regimes, when documented properly, can show differing visions at a granular level (Goldstein Reference Goldstein2020). Second, as schools are one of the central, earliest, and most sustained pathways through which the state interacts with its citizens, it is especially in the area of education that we might expect state-promoted narratives to shape public understandings of national identity. Third, these stylized facts can generally be applied to many if not all states, which widely advocate for their preferred vision of history through some form of official school curriculum.
In India, textbooks have been targeted since at least the 1990s both by the secular left and the Hindu right, with each accusing the other of re-writing content when in power to advance political aims. Textbooks promoted by the Hindu right “construct a particular imagined community: one defined according to Hindu cultural similarity,” while those advanced by the secular left aim to “promote the building of a secular national identity” (Guichard Reference Guichard2010, 56). Debates regarding textbook content thus mirror the larger ideological “battle over history” which “pits a Nehruvian narrative of India’s past, which stresses plurality, complexity, and tension, against a Hindutva narrative, which stresses internal purity and external danger” (Nussbaum Reference Nussbaum2009, 261). Changes to textbooks produced by the National Council for Education Research and Training (NCERT)—a federal institution in charge of developing curricula and preparing model textbooks—reflect the push and pull of the “education war” era: textbooks written in the 1990s under INC were replaced in 2002 by a BJP-led coalition, only to be changed in 2006 and 2008 when INC took back control, and heavily modified once again when BJP returned to power in 2014 (Guichard Reference Guichard2010). Similar patterns are visible with state-specific textbooks, which have been markedly revised in recent years alone (Traub Reference Traub2018).
We use excerpts from official history textbooks as our experimental treatments. The selected passages provide clear, state-sponsored narratives corresponding to the two main historical visions summarized above. Further, because our treatments reflect real-world content, we connect them to school, and we identify the state as their source, we are more confident in their ecological validity.Footnote 10 Our treatments do not fully approximate how individuals engage with textbook material, but we argue that they do so better than available alternative approaches, for instance, if we had simply invented our own content and labeled it “hypothetical” textbook material.Footnote 11 Lastly, while we aim to maximize the ecological validity of our textbook treatments, we note that our ultimate focus in this paper concerns the effects of narratives—not the medium through which they are conveyed.Footnote 12
Religion and Leadership
Muslims, who constitute India’s largest minority group at approximately 14% of the country’s population and two hundred million people, are heavily underrepresented in the country’s leadership positions (Adeney and Swenden Reference Adeney and Swenden2019; Jaffrelot Reference Jaffrelot2021; Khan Reference Khan2020). They are estimated to make up under 3% of directors and senior executives at the country’s largest five hundred companies, only 5% of its government sector employees and an even smaller portion of its civil servants, disproportionately small fractions of the labor force and employers and employees in key economic sectors, and a small fraction of its representatives in political institutions—including under 5% of parliamentarians (Adeney and Swenden Reference Adeney and Swenden2019; Jaffrelot Reference Jaffrelot2021; Khan Reference Khan2020). Among minority groups, Muslims are indeed the most underrepresented in many of these institutions and organizations, with trends pointing toward their deepening exclusion (Adeney and Swenden Reference Adeney and Swenden2019; Jaffrelot Reference Jaffrelot2021; Khan Reference Khan2020).
We believe that our theory is of import and relevance to Muslim representation for a few reasons. First, many of the leadership positions in which Muslims are underrepresented—as business executives, trade union leaders, bureaucrats, newspaper editors, and politicians—require speaking on behalf both of marginalized and dominant group members. Second, many of these positions, including ones in economic and civil society organizations, have been found to be critical to political outcomes in India. For instance, Varshney (Reference Varshney2003) demonstrates that Hindu–Muslim violence was less likely where intercommunal civic networks were stronger. Towns with a high degree of “interreligious interaction in nondenominational organizations” ranging from art to trade associations—and with religious diversity in their leadership—developed peace committees and media campaigns that prevented rioting (Varshney Reference Varshney2003, 127).
Third, scholars have tied the rise of Hindu nationalism to increasing Muslim underrepresentation across a wide range of positions, in sectors ranging from media organizations to business and politics, and have noted the close relationship between national-level debates and local-level political outcomes (Adeney and Swenden Reference Adeney and Swenden2019; Jaffrelot Reference Jaffrelot, Shastri and Wilson2001; Khan Reference Khan2020; Varshney Reference Varshney2003).Footnote 13 Fourth, scholars have highlighted the importance of both demand- and supply-side factors in informing representational outcomes in India (Auerbach et al. Reference Auerbach, Bussell, Chauchard, Jensenius, Nellis, Schneider and Sircar2022; Goyal Reference Goyal2020).
RESEARCH DESIGN
We test our theory through an incentivized online experiment conducted with 1,592 Indian Muslim and Hindu respondents between June and August 2021. We randomly assign participants to either complete a history exercise with an exclusive, inclusive, or a “neutral” (baseline) historical narrative.Footnote 14 Our main outcomes are the perceived centrality of Hindus and Muslims to the nation, as well as the perceived entitlement of, supply of, and demand for members of the Muslim minority in leadership positions.
Sample
Participants were recruited using an online panel from the market research company Lucid, which creates its panel by aggregating across a number of survey panels. Lucid is increasingly popular with social scientists, and validation efforts from the United States indicate that respondents recruited on the platform approximate national benchmarks, both demographically and behaviorally (Coppock and McClellan Reference Coppock and McClellan2019). To ensure high data quality, we use pre-treatment attention checks suggested by Aronow et al. (Reference Aronow, Kalla, Orr and Ternovski2020) in a recent paper on inattentiveness among Lucid survey respondents (see Section G.4 of the Supplementary Material), and we exclude respondents who take the survey very quickly (i.e., less than half the median soft launch completion time) or very slowly (i.e., more than twice the median time).Footnote 15
Our sample includes respondents from four Indian states: Delhi, Gujarat, Maharashtra, and Uttar Pradesh. India is an enormous, diverse country, and restricting sampling to a few states allows us to maintain some control over the sociopolitical contexts in which respondents are embedded. Section G.3 of the Supplementary Material describes the criteria we used when selecting these four states in detail. We screen out non-Muslims and non-Hindus, and we set quotas for each religious identity category as well as for men and women. Participants could complete the study in English or Hindi and receive a participation fee, along with additional earnings based on their decisions, the decisions of others, and chance, in the form of e-rewards/e-currency. Our sample—which was more urban and educated than the general population—appears similar to other online samples of Indian respondents recruited through a variety of platforms such as Facebook, MTurk, and Qualtrics (Boas, Christenson, and Glick Reference Boas, Christenson and Glick2020; Guess et al. Reference Guess, Lerner, Lyons, Montgomery, Nyhan, Reifler and Sircar2020). While our sample is seemingly not an outlier as regards other online samples collected in India, behavior in online samples—an important demographic in their own right—often, but not always, generalizes to the broader population. Our results may thus be best interpreted as speaking to Indian Internet users, who more often reside in urban areas and are more educated.
Study Stages, Experiment Treatments, and Measurement
Figure 1 summarizes the four stages of our study (see Section F of the Supplementary Material for experiment materials).
Sequence of Study Stages
Stage 1
In the first stage, participants complete a short demographic questionnaire, at the end of which they are asked to choose an avatar, which is visibly Hindu or Muslim, to represent them at a later stage in the study. Respondents are given a choice between three gender and religion congruent avatars, which we selected from a larger set due to their popularity with respondents and their high degree of religious identifiability in a pre-test (
$ N=189 $
) we conducted in October and November 2020. To reduce the likelihood that respondents might have a negative reaction or feel that they were forced into being represented by a religious stereotype, we included some variation in avatars’ style of dress (in particular, modern versus traditional); importantly, we found in the pre-test that respondents overwhelmingly chose identity-congruent avatars when also given the option of identity-incongruent ones, and did not voice any concerns with avatars or the selection task (see Section D.1 of the Supplementary Material).
Stage 2
In the second stage, participants are informed that the theme of the study is “Back to School.” They are randomly assigned a history text, sourced from real official school textbook material, that either highlights the plural nature of Indian history and Muslims’ historical contributions to the nation (inclusive representation treatment), describes a glorious Hindu past and a dark period of Muslim invasions (exclusive representation treatment), or which details history on non-politicized topics such as agriculture (baseline condition). Respondents are informed that the material comes from textbooks “approved by government educational boards in India.” Passages are followed by comprehension questions designed to highlight the critical elements of the texts and for which correct answers contribute to participants’ payoffs.
When identifying young adult textbook material using the websites of the NCERT and state educational boards, we tried to choose passages that clearly illustrated the strategies different sides pursue to shape their preferred notion of nationhood, as summarized in Table 2 and further detailed in Section E.2 of the Supplementary Material (Guichard Reference Guichard2010; Khan et al. Reference Khan, Svensson, Jogdand and Liu2017; Nussbaum Reference Nussbaum2009). In particular, we focused on two periods, ancient and medieval India, that constitute critical historical moments and are interpreted very differently by the two competing narratives. For the Hindu right, Ancient India provides an opportunity to establish a Hindu “foundational myth” (Khan et al. Reference Khan, Svensson, Jogdand and Liu2017). According to this account, Hindus are the direct descendants of the indigenous Indo-Aryans, who—prior to Muslim invasions and colonial conquest—enjoyed a “golden age” of peace, harmony, cultural and scientific innovation, and prosperity (Guichard Reference Guichard2010). Medieval India, on the other hand, is described as a dark period of Muslim invasion, destruction, violence, and Hindu subjugation.
Historical Representations
The secular left offers a radically different version of Indian history in which Ancient India was characterized by significant heterogeneity and the coexistence of multiple religions. According to this version, the Mughal rulers of Medieval India positively contributed to the nation through their embrace of syncretic culture, contributions to the arts (poetry, architecture, and cuisine), techniques of governance, and tradition of religious tolerance. Table 2 summarizes key characteristics and events associated with these two time periods according to the secular (inclusive representation treatment) and the Hindu nationalist (exclusive representation treatment) versions of history, and as detailed in our experimental treatments.Footnote 16
To increase the salience of our treatments and maintain it through later stages of the study, as well as to make the study appear more natural to respondents by highlighting the topic matter, we additionally include as a banner content that a student might see in a classroom and which corresponds to our treatment material. Thus, in the exclusive treatment, subjects see images of Hindu historical figures on every page beginning in Stage 2, whereas in the inclusive treatment subjects see images of Mughal art. These images also come from real history textbooks (see Figure F19 in the Supplementary Material).
Following the texts and comprehension questions, respondents answer two questions designed to measure the perceived centrality (prototypicality and historical contributions) of Hindus and Muslims to the nation (
$ {H}_1 $
). To measure perceived contributions, we ask respondents to select from a list which groups they believe most contributed in a positive way to the development of the Indian nation. Participants are asked to pick a total of five words from a list of 19. In our analysis, we count how many of the selected words are distinctly “Hindu” or “Muslim” in nature, with variables therefore ranging from 0 to 5 where “0” indicates that no Hindu/Muslim words were selected and “5” indicates that all words selected were Hindu/Muslim in nature. To measure prototypicality, we ask respondents to what extent they agree with the following two statements: (1) “I think it would be accurate if someone described me as a typical Indian” and (2) “I feel similar to other Indians.” The order in which the two questions appear is randomized.
Stage 3
Following Stage 2, participants are matched in groups of three and shown the religiously-identifiable avatars of their partners, who they are told were drawn “from a pool of participants that was chosen be broadly representative of the Indian nation (though certain communities may appear more or less than others).”Footnote 17 They learn that they must complete two additional school exercises on geography and sociology, but only one of them will count for payment.Footnote 18 Payoffs are determined by group performance and all members will earn an equal share: 1,000 Indian rupees (INR) for each correct answer.Footnote 19 While respondents complete both additional exercises, a group representative will pick which of the two exercises will count for payment for all group members. We elicit individuals’ desire to be group representative by asking them to state from 1 to 4 their willingness to be group representative. Following Coffman (Reference Coffman2014), we explain this decision in terms of reserving a “place in line” relative to one’s group partners, where “1” means that a respondent wishes to be first in line to be group representative and “4” means they do not even wish to be in line. After stating their number, participants are asked to rank from 1 to 3 (including themselves) their preference for which group member should be group representative, which they are told will be determinative in the event that multiple group members are equally most willing to be group representative.
The incentivized group task yields two outcomes that we use to evaluate the supply of and demand for different types of leaders (
$ {H}_3 $
–
$ {H}_5 $
), with leadership operationalized as taking a decision that will affect others (in a group broadly representative of the Indian nation): one that measures participants’ willingness to become group representative (supply) and a second that captures participants’ willingness to see other group members as representative (demand).Footnote
20 We expect these outcomes to be less vulnerable to social desirability bias than are stated preference questions, both because it is less clear what a socially desirable answer is and because the decisions are financially incentivized; they also offer a relatively high degree of experimenter control, as differences in interpretations and predispositions are less likely than in an area over which individuals already have established opinions or views.
While the “lab-like” group task thus offers some distinctive benefits, we also collect supplementary revealed and stated preference outcomes that relate more directly to real-world behavior of import. To capture our supply-side channel, we explore the extent to which the different narratives affect participants’ willingness to seek out information on how to become engaged in politics. After informing respondents that recent studies on leadership and political engagement in India indicate that many people lack information about how to become politically active, we ask whether they would like to receive information on platforms and programs that work toward empowering citizens to participate in democratic and governance processes. Later, respondents who answer “yes” to this question receive a link to a website where, after entering their unique ID number, they are shown the promised information.
In order to further capture the demand side, we explore how exposure to the different representations of history affect evaluations of real Hindu and Muslim politicians. Several NGOs in India track politicians’ performance and publish “report cards” in order to improve accountability. We randomly assign respondents a report card issued by an NGO in Maharashtra for either a Hindu or a Muslim politician with a similar rating.Footnote 21 Respondents are asked to rate the politician on several different dimensions, such as qualification and deservingness for the office and how representative they are of India.
In Stage 3, we also measure perceived entitlement to speak on behalf of others in the nation (
$ {H}_2 $
) in two ways. First, we use two criteria on which respondents were asked to evaluate the Muslim politician in the report card exercise: perceived deservingness and qualification for office. Second, we ask respondents to share whether they feel (1) worthy and (2) qualified to be a group leader in the group task. These measures correspond closely to our conceptualization of entitlement as reflecting a person’s perceived deservingness as well as their perceived ability to be a group representative, by virtue of who they are. All item answers are elicited on 5-point agree-disagree scales. See Section F.3.3 of the Supplementary Material for more details on Stage 3 measurement.
Stage 4
In the study’s final stage, participants complete the geography and the sociology exercises and, subsequently, receive information about payment. To simplify our design, we allocate participants to groups, and calculate payments, ex post; we implement quotas for Muslims and Hindus so that there is no deception and individuals are assigned to groups with the members they were shown during the study. These design choices free us from having to conduct the study with simultaneous decision-making, which due to attrition and other challenges associated with an online experiment, would be challenging. Participants receive their participation fee and are given the details of a webpage where, after all participants have completed the study, they can redeem any additional gift card amount earned by entering a unique survey code. The experiment is thus entirely anonymous and does not require the collection of participants’ personal information at any point.
EMPIRICAL STRATEGY AND ETHICS
Empirical Strategy
In our main analysis, we run a series of OLS regressions where we investigate effects of our exclusive and inclusive treatments (relative to the baseline), both for the overall respondent population, and—as is most relevant for evaluating our hypotheses—separately for Muslim and Hindu respondents. To evaluate
$ {H}_5 $
, we consider treatment effects conditional on the religious composition of a participant’s randomly assigned group. See Section G.1 of the Supplementary Material for our pre-specified model specifications.
In order to evaluate the study design and to inform our power analysis, we conducted pilot sessions between December 2020 and March 2021 (
$ N=402 $
). Using means and standard deviations from the pilot and from a study on a different subject that used a similar group task outcome (Coffman Reference Coffman2014), our estimated sample size varied from as little as
$ N\approx $
400 respondents to as many as
$ N\approx $
1,100, depending on the data source and outcome (see Section G.2 of the Supplementary Material). We took a conservative approach and sampled
$ N=1,592 $
respondents.
Ethics
Given the sensitive nature of the topic under study, we took several steps to ensure that our design met a high ethical standard. First, in a post-experiment debrief, we explained to participants the purpose of our study and noted that the depictions they were shown are contested and should not be taken as fact. Second, we shared our pre-analysis plan with Muslim and Hindu faculty at Indian institutions to receive their input on the study’s ethics. Third, we chose to use treatment material sourced from real Indian textbooks not only to further the study’s ecological validity, but also because we felt it would increase the likelihood that participants had seen similar material in the real world. Fourth, in addition to probing possible negative effects of exclusive representations, our study also considered whether inclusive representations can have positive impacts. The study received IRB approval from the ethical board at Aarhus University.
We discuss these steps and our reasoning at greater length in Section G.6 of the Supplementary Material. We also present evidence from a pre-test (
$ N=189 $
, October–November 2020) and our full study indicating that participants did not find our experimental treatments upsetting. We find that respondents overwhelmingly report positive perceptions of the study and say that treatment material was similar to the types of lessons they might have learned at school; importantly, we do not detect statistically distinguishable differences by treatment text (inclusive, exclusive, or control) or respondent religion (Hindu or Muslim). We thus believe that our research design, including our exclusive historical treatment text, was unlikely to expose participants to psychological costs higher than those they might experience in real life.
RESULTS
We organize the presentation of results around our stated hypotheses. We thus begin with an analysis of perceived centrality to the nation, before considering perceived entitlement to lead, the demand for and supply of different types of leaders, and the effects of different group compositions. Lastly, we evaluate our assumptions and interpretation of results using participants’ open-ended responses. Our sample contains
$ N=839 $
Hindus (53%) and
$ N=753\hskip2pt (47\%) $
Muslims. We present coefficient plots and point estimates of the mean levels of our outcomes of interest. Tables reporting regression model output can be found in Section C of the Supplementary Material.
Perceived Centrality to the Nation
Our first hypothesis states that historical representations shape the perceived centrality (historical contributions and prototypicality) of different groups to the nation, with exclusive narratives reducing the perceived centrality of marginalized groups (
$ {H}_{1a} $
) and inclusive narratives increasing their perceived centrality (
$ {H}_{1b} $
). Figure 2 shows effects of our inclusive and exclusive treatments (relative to the baseline), overall and split by respondent religion. We consider as outcomes perceived relative historical contributions of Muslims (top left panel), perceived contributions of Muslims (top right panel), perceived contributions of Hindus (bottom left panel), and individuals’ self-assessed prototypicality (averaged over two survey items on similarity to others and a typical Indian).
Perceived Centrality
Note: Figure shows treatment effects of historical representations on relative perceived historical contributions of Muslims versus Hindus (panel a), perceived contributions of Muslims (panel b), perceived contributions of Hindus (panel c), as well as on respondents’ self-assessed prototypicality (panel d; prototypicality is the average of the two survey items). We display both overall results, and results split by respondent religion. Here, 95% (thin line) and 90% (thick line) confidence intervals are shown. See Table B5 in the Supplementary Material for full regression model output.
Results are largely supportive of
$ {H}_1 $
. Beginning with the top left panel, we observe that the exclusive treatment results in a decrease in the relative perceived centrality of Muslims (
$ {H}_{1a} $
; 95% confidence interval (CI) [−0.59,−0.16]), whereas the inclusive treatment results in an increase in their relative perceived centrality (
$ {H}_{1b} $
; 95% CI [0.01, 0.43]). Shifts in perceptions are starkest where they reflect a positive shift in the relative centrality of an individual’s group: overall increases in relative Muslim centrality under the inclusive treatment are driven by Muslim respondents, and overall increases in relative Hindu centrality under the exclusive treatment are driven by Hindu participants. However, as the panels on perceived Muslim (top right) and Hindu (bottom left) contributions demonstrate, both groups respond to the treatments, evaluating Muslim centrality as higher under the inclusive treatment and Hindu centrality as higher under the exclusive treatment. Effect sizes are often of substantial magnitude: the number of Hindu groups chosen by Hindu respondents, for instance, jumps from an average of 1.76 in the baseline treatment to an average of 2.33 in the exclusive treatment, an increase of nearly a third.Footnote
22
While our findings indicate that our treatments affected perceived contributions in the expected direction, we find little evidence for corresponding shifts in perceived prototypicality of different groups to the nation (bottom right panel). While we can only speculate as to the reasons for this difference, one could be the framing of our questions: whereas our contributions question asked explicitly about perceived contributions of different groups to the nation, our prototypicality questions asked respondents to evaluate their own prototypicality to the nation. Explicit self-assessments might be more difficult to shift in the short-term, and might not fully capture the idea of perceived group-level prototypicality; the personal nature of the questions also may have provoked unexpected responses. An item capturing more implicit associations of groups and the Indian nation may very well have produced different results.Footnote 23
Perceived Entitlement to Lead
According to our second hypothesis, historical representations will shape perceived entitlement of different groups in the nation, in particular, by reducing marginalized group members’ perceived entitlement where a representation is exclusive (
$ {H}_{2a} $
) and increasing their perceived entitlement where a representation is inclusive (
$ {H}_{2b} $
). Figure 3 shows results, with the top row corresponding to individuals’ evaluations of whether the Muslim MLA politician shown in the report card is qualified and deserving of office, and the bottom row corresponding to participants’ self-assessments of their own qualifications and worthiness of serving as group representative. As before, we display treatment effects overall and by respondent religion.
Perceived Entitlement
Note: Figure shows treatment effects of historical representations on whether individuals perceived the Muslim MLA shown in the report card as qualified (panel a) or deserving (panel b) of office, as well as whether individuals perceived themselves as qualified (panel c) or deserving (panel d) of serving as group leader. We display both overall results, and results split by respondent religion. Here, 95% (thin line) and 90% (thick line) confidence intervals are shown. See Table B6 in the Supplementary Material for full regression model output.
We find weak support for
$ {H}_2 $
. In particular, we observe that Muslim respondents in the inclusive treatment are significantly more likely to evaluate the Muslim MLA as qualified for office (
$ {H}_{2b} $
; 95% CI [0.04, 0.42]) and deserving of office (
$ {H}_{2b} $
; 90% CI [−0.00, 0.31]). We also observe that Muslim respondents report feeling more deserving (“worthy”) to act as group representative, though this difference—an increase from 4.42 on average in the baseline condition to 4.53 in the inclusive condition—is not statistically distinguishable from zero at conventional levels (
$ p=0.15 $
). Lastly, Hindu respondents become less likely to view the Muslim MLA as qualified in the inclusive treatment (
$ p=0.08 $
), a shift potentially consistent with backlash (though results overall are more consistent with a null effect, see our “Discussion” section).Footnote
24
Who Becomes Leader? Considering Effects on Supply and Demand
Thus far, we have found that historical representations can affect perceived historical contributions of different groups to society (their “centrality”) as well as their perceived qualifications (their “entitlement”) to lead. Next, we consider whether and how historical representations affect the supply of and demand for different types of leaders (
$ {H}_3 $
and
$ {H}_4 $
), and to what degree patterns are conditional on the composition of the group one would be tasked with leading (
$ {H}_5 $
). We first evaluate effects on leadership using our incentivized group task outcomes, before turning to additional questions with more explicit ties to real-world behavior.
Incentivized Behavior in the “Lab”
Supply of Leadership.
Beginning first with the supply-side,
$ {H}_{3a} $
predicted that exclusive historical narratives should decrease the willingness of marginalized group members to assume a leadership role, whereas
$ {H}_{4a} $
predicted that inclusive narratives should increase their willingness. Results are reported in Figure 4. We find strong support for
$ {H}_{4a} $
: exposure to an inclusive narrative significantly increases Muslim respondents’ willingness to volunteer as group representative (95% CI [0.06, 0.27]). Muslims in the inclusive treatment are even more willing to lead than are Hindus.Footnote
25 Results are not consistent with history as a “zero sum” game: while greater historical inclusion of marginalized groups increases the supply of potential leaders from those communities (Muslims), it does not have any discernible effect on the supply of the dominant group (Hindus). Indeed, we observe that the increase in Muslims’ willingness to volunteer as representative is sufficiently large that it results in an overall increase in respondent willingness to lead (95% CI [0.00, 0.15]).Footnote
26
Treatment Effects on Supply of Leadership
Note: Figure displays treatment effects of historical representations on willingness to take on a leadership role, with panel a showing means with 95% intervals and panel b showing output from linear regression models excluding controls (bright colors) and including sociodemographic controls (dim colors). The outcome variable ranges from 1 (not at all willing) to 4 (very willing). See Table B2 in the Supplementary Material for full regression model output.
While the inclusive history treatment positively affects the supply of minority leadership, we do not observe the expected outcomes among those exposed to exclusive renderings of history; the exclusive history treatment has a small, positive, and statistically indistinguishable from zero effect on minority members’ willingness to lead. Although it is unclear why we do not observe a negative effect of exclusive history (
$ {H}_{3a} $
), as we elaborate on in our “Discussion” section, one potential explanation could be that the exclusive historical narrative is currently very salient in Indian discourse and therefore does not have an equally powerful impact on respondents’ attitudes and behavior.
Demand for Leadership.
Hypotheses
$ {H}_{3b} $
and
$ {H}_{4b} $
concern the role of historical representations in shaping demand for Muslim group leaders. We posited that exclusive renderings of history would negatively affect demand for Muslim leaders, and inclusive renderings would positively affect demand. As described, participants were asked to rank their group members according to how much they would like to see each of them as group leader in the event of a tie. We examine demand for Muslim group leaders by observing how Hindus who were randomly matched with one Hindu and one Muslim partner (in their group of 3) rank their Muslim group member.
The panel on the left of Figure 5 shows how Hindus in Hindu-majority groups ranked their partners. As can be observed, Hindu respondents rank their Hindu partners higher than their Muslim partners in this exercise. Whereas Hindu respondents rank their Hindu partner last 37% of the time, they rank their Muslim partner last 53% of the time.Footnote 27 However, as is clear from the right panel, we do not find any evidence that our treatments affect this apparent pro-Hindu bias. Treatment effects are indistinguishable from zero, and if anything, point in the opposite direction of what was expected: point estimates of pro-Hindu bias are larger in the inclusive treatment group, and smaller in the exclusive treatment group, as compared with the baseline condition.Footnote 28 , Footnote 29
Demand for Muslim Leaders among Hindu Respondents
Note: Panel a shows how Hindus in Hindu-majority groups ranked their Hindu and Muslim group members. The values indicate the proportion of Hindus who ranked their group member as first, second, or last. Panel b shows expected pro-Hindu bias in the different treatment conditions which we calculate by subtracting the rank given to a Muslim partner from the rank given to a Hindu partner. The resulting variable ranges from −2 to 2, with more positive values indicating a stronger preference for Hindu partners. Here, 95% confidence intervals are shown. See Table B3 in the Supplementary Material for full regression model output.
Group Composition.
According to our theory and the results presented thus far, historical representations can affect perceived belonging to the nation and entitlement to speak on the behalf of its “true” members. Building on this logic, our fifth hypothesis posits that the effects of historical representations on marginalized group members’ willingness to seek out leadership roles will be greater when the role requires that marginalized group members speak on behalf of unquestioned “core” members of the nation, than where the position requires that they speak on behalf only of other marginalized group members. To evaluate this hypothesis, we randomly assigned participants to either religiously homogeneous groups or religiously mixed groups. We expect the effects of both exclusive and inclusive historical narratives to be stronger where group compositions are mixed.
Figure 6 shows the relationship between group composition, our history treatments, and Muslim respondents’ willingness to lead. The left panel displays regression output where we interact treatment assignment with either an indicator for whether one was randomly assigned to a religiously mixed or homogeneous group (“binary” column), or with a categorical variable that additionally distinguishes between Muslim-majority and Muslim-minority mixed groups (“full distribution” column). Consistent with
$ {H}_5 $
, we can see in the “binary” column that Muslim respondents are particularly likely to supply leadership when they are assigned both to the inclusive treatment and a religiously mixed group, a result that is statistically distinguishable from zero at a 10% level. Interestingly, the “full distribution” column shows that this effect is driven by mixed Muslim majority groups (
$ p=0.01 $
), which may reflect a higher threshold to increasing minority individuals’ perceived entitlement to lead where the majority group constitutes a larger portion of the population under the leader’s authority. We do not find much evidence of an interaction effect with the exclusive treatment, though we do observe that Muslim respondents are less willing to lead where they are randomly assigned both to the exclusive treatment and a Muslim minority group (
$ p=0.08 $
).
Group Composition and Willingness to Lead among Muslims
Note: Panel a displays regression output where we interact treatment assignment with a binary (“binary” column) or categorical (“full distribution”) indicator for the religious composition of a participant’s randomly assigned group (homogeneous, mixed Hindu majority, or mixed Muslim majority). Here, 95% (thin line) and 90% (thick line) confidence intervals are shown. Panel b shows simulated sampling distributions for respondents in the control and inclusive treatment conditions, by their group composition. We simulate sampling distributions by taking one thousand random draws from a multivariate normal distribution with means corresponding to the coefficients and variance corresponding to the variance–covariance matrix. The outcome in both cases is willingness to lead and analysis is limited to Muslims. See Table B4 in the Supplementary Material for full regression model output.
We further investigate the relationship between the inclusive treatment and group composition in the right panel of Figure 6. Specifically, we use results from our study to simulate sampling distributions of the expected willingness to lead among Muslim respondents in the control and inclusive treatment conditions and by group composition. We see that exposure to inclusive history does not increase willingness to lead among Muslim respondents in groups where all members are Muslim. In mixed groups, however, a clear difference emerges: respondents in the inclusive history treatment are significantly more willing to volunteer as group representative than are respondents in the control condition. The difference between inclusive treatment and control increases from 0.05 (95% CI [−0.10; 0.21]) in homogeneous groups to 0.24 (95% CI [0.11, 0.36]) in mixed groups. This lends additional support for
$ {H}_5 $
.
Open-Ended Responses.
Answers to follow-up questions on why participants elected to serve as group representative or not are consistent with a number of our assumptions and interpretations. First, they indicate that participants viewed the group representative position as a leadership role. A number of respondents said that they wanted to be representative because they viewed themselves as leaders; the comment by one participant that they wished to be representative because “I have leadership qualities” was a common refrain. Second, in justifying their qualification for the role, participants often stated their deservingness and worthiness for the role—matching exactly our conceptualization of perceived entitlement to lead.Footnote 30 One participant wrote “I am qualified to be group leader and I am worthy of being group leader.” Another said they wanted to be group leader “because I think I truly deserve it.” A lack of perceived entitlement also underlied responses for why individuals did not want to be group leader; as one respondent wrote, “Because I’m not sure if I’m worthy of it or not.”
Third, there were indications that respondents viewed entitlement to serve as group representative as related to the Indian nation, and their perceived belonging within it. One participant said that they wanted to be group leader because “as an Indian, I want to represent my country.” Certain individuals indicated that they felt that their social group had a particularly strong claim to the nation’s core community. One wrote that they wanted to be representative because “my group represents India and Indian religions.” And multiple respondents said that they wanted to serve as leader simply “because I am Hindu.”Footnote 31
Outcomes with Real-World Political Corollaries
In addition to the incentivized game, we also measure outcomes relating more explicitly to real-world behavior of import. To capture the supply-side channel, we examine how the history treatments affect respondents’ willingness to seek out information on citizenship participation; to capture the demand-side channel, we assess how exposure to different representations of history affect evaluations of real Hindu and Muslim politicians.
Supply of Leadership (Willingness to Seek Out Information on Citizen Participation).
To measure the supply of leadership, we offer participants information on how to become politically involved and provide a website at the end of the study to those requesting information. Figure 7 shows treatment effects on the likelihood that respondents elected to receive the information. We find that the exclusive treatment has a small, negative effect on Muslims’ willingness to seek out political information (95% CI [−0.08,−0.00]); while 95% of Muslim respondents request information in the control condition, this percentage declines to 91% in the exclusive treatment. While political interest is very high in our sample overall, the findings nevertheless lend support to the theoretical argument that exclusive renderings of history could reduce political participation among marginalized group members by discouraging them from taking the first necessary steps in the process of becoming politically engaged (
$ {H}_{3a} $
). However, we find no corresponding effect with inclusive representations and willingness to receive information, potentially because willingness is very high at the baseline.Footnote
32
Treatment Effects on Willingness to Seek Out Information
Note: Figure displays treatment effects of historical representations on willingness to receive information on how to become politically involved (dummy 0–1). Panel a shows means with 95% intervals and Panel b shows output from linear regression models excluding (bright colors) and including (dim colors) sociodemographic controls. See Table B7 in the Supplementary Material for full regression model output.
Demand for Leadership (Evaluation of Politicians).
As an additional measure of demand, we randomly assigned participants report cards for either a real Hindu or Muslim politician and asked them to evaluate the politician on seven different dimensions. These dimensions include whether the politician is perceived as qualified, deserving of office, representative of his constituents, representative of India, popular, typical (similar to other politicians) and whether respondents report feeling comfortable being represented by this politician. The Hindu and Muslim politicians in this exercise both have a similar rating (see Section F.3.3 of the Supplementary Material for details). Using inverse covariance weighting (Anderson Reference Anderson2008), we create an index (“overall evaluation”) capturing to what extent respondents evaluated a politician positively.
As shown in Figure 8, the history treatments did not affect how respondents evaluated the Hindu politician. However, we do observe a significant increase in positive evaluations of the Muslim politician among Muslim respondents assigned to the inclusive treatment (95% CI [0.02, 0.30]): being exposed to an inclusive history enhances ratings of the Muslim candidate from an overall evaluation of 4.22 (95% CI [4.12; 4.31]) to 4.37 (95% CI [4.28; 4.47]) on a 1–5 scale. These results offer additional support for
$ {H}_{4b} $
.Footnote
33
Treatment Effects on Politician Evaluations
Note: Figure displays treatment effects of historical representations on evaluations of Hindu and Muslim politicians (MLAs). Coefficients are based on linear regression models excluding controls (bright colors) and including sociodemographic controls (dim colors). Here, 95% (thin line) and 90% (thick line) confidence intervals are shown. The outcome variable ranges from 1 to 5. See Table B8 in the Supplementary Material for full regression model output.
DISCUSSION
Thus far, we have presented evidence consistent with a number of our theoretical expectations. We observed that historical representations can affect the perceived centrality of different groups to the nation (
$ {H}_1 $
), found inconsistent evidence that representations can affect perceived entitlement to lead (
$ {H}_2 $
), and saw support for our expectation that representations can affect the supply of and demand for marginalized group members in leadership positions (
$ {H}_3 $
and
$ {H}_4 $
)—particularly those that demand leadership over ethnically heterogeneous groups (
$ {H}_5 $
). We find the strongest support for our theory when considering the inclusive treatment among Muslim respondents, which led to increases in the perceived centrality and entitlement of Muslims, increases in their supply of leadership (especially in mixed groups), and increases in demand for their leadership (as measured by evaluations of Muslim politicians). We observe lesser effects among Hindus and in the exclusive treatment—which, while it increases the perceived centrality of Hindus and diminishes the supply of leadership (as measured by willingness to seek out information on citizenship participation) among Muslim respondents, otherwise has little discernible impact.
Interpreting Unanticipated Findings
Why did we not observe stronger effects from our exclusive treatment? We find support for multiple possible interpretations. We speculate that one potential explanation for the comparatively weak effects of our exclusive treatment concerns the status quo. Specifically, people in India have recently been inundated with exclusive historical narratives. As we have previously noted, history has been an active battlefield in India for some time, and an ambitious revisionist campaign is well underway.Footnote 34 If people had already been exposed to exclusive narratives, then our exclusive treatment may have not provided sufficient new information for them to update their beliefs or perceptions.Footnote 35 If true, then we would also expect inclusive narratives that differ from many status quo teachings to have a greater impact, which is what we observe.Footnote 36
Why did we not observe stronger effects of our inclusive treatment on demand among Hindus? Experimental evidence is more consistent with the inclusive treatment having a null effect among Hindus, as opposed to provoking backlash: while patterns in a few cases point toward Hindu respondents responding negatively to the treatment (see Figures 3 and 5), differences are not statistically distinguishable from zero at conventional levels; for the majority of outcomes, the point estimate approximates zero. Further, we observe both few Hindus overall registering very negative reactions to the inclusive treatment, as well as little difference in this proportion between treatments (see Figure C4 in the Supplementary Material).
Alternative Explanations
In Section C of the Supplementary Material, we consider some alternative explanations for our findings. In particular, we evaluate alternative explanations for one of our central findings that inclusive renderings of history increase marginalized group members’ willingness to take on leadership positions. Three possible alternative mechanisms come to mind. First, inclusive renderings of history may simply improve Muslim respondents’ overall experience with—and engagement in—the survey. Put differently, Muslim respondents may choose to volunteer as group representative not because they feel more entitled to take decisions on behalf of others, but because they are enjoying the survey and are more eager to participate actively in it. Second, Muslim respondents may score better in the inclusive history exercise because they are more familiar with the material and/or find the correct responses to be more consistent with their worldview. If this were the case, Muslim minorities may be more likely to volunteer because they believe more strongly in their performance in the exercises. Third, participants may be inferring that the authors of the study hold the view advanced by the narrative they are presented with, and therefore are more likely to give the answers they think the study’s authors want to hear (i.e., Muslim respondents volunteer more due to demand effects).
In order to examine the first two alternative mechanisms, in Table C16 in Section C of the Supplementary Material, we examine how the history treatments affect firstly overall engagement with the study measured through an item asking respondents to what extent they were involved with the study, and secondly respondents’ score in the history exercises. The findings reveal that the inclusive history treatment does not increase Muslim respondents’ overall engagement with the study. Muslim respondents also did not score better in the inclusive history exercises. In light of this evidence, it seems unlikely that engagement or performance could better explain why minorities volunteer more when exposed to inclusive history. In order to explore the third alternative mechanism, we evaluate different observable implications derived from the argument about demand effects in Section C.4 of the Supplementary Material. Overall, we find little support for the notion that strategic motivations related to demand effects explain our findings.
Revisiting Implications for Representation Outcomes
We theorize that inclusive narratives can increase the descriptive representation of marginalized individuals by increasing both their willingness to seek out leadership roles (supply channel) and of greater society to grant them such roles (demand channel). As discussed above, however, while we find evidence consistent with inclusive narratives possibly increasing supply in the form of Muslims’ leadership ambitions, we observe little support for our expectation that such narratives would increase Hindu demand for Muslim leadership. Given observed movement among Muslim but not Hindu respondents, what are the possible implications of our findings for representation outcomes?
In answering this question, it is important to first contextualize ambition to lead as one, albeit critical, component of a longer leadership “pipeline.”Footnote 37 For example, in politics, one might imagine this pipeline beginning with a pool of candidates eligible to run for office (“potential aspirants”) and ending with the subset elected for office (“elected aspirants”), with a number of steps—and obstacles—contained therein. Through this lens, we might interpret our treatment as increasing the proportion of Muslim potential aspirants who develop “nascent ambition,” or “a professed desire to hold office” (Bernhard, Shames, and Teele Reference Bernhard, Shames and Teele2021, 380). A large body of literature testifies to the importance of increasing marginalized individuals’ leadership ambitions and argues that doing so can carry significant implications for marginalized group descriptive representation (Bos et al. Reference Bos, Greenlee, Holman, Oxley and Celeste Jay2022; Clayton, O’Brien, and Piscopo Reference Clayton, O’Brien and Piscopo2023; Fox and Lawless Reference Fox and Lawless2005). Indeed, even those scholars who argue that more attention should be paid to structural, demand-side factors typically recognize that increases in marginalized groups’ nascent ambition are necessary, even if not sufficient, to efforts to increase their levels of descriptive representation (Bernhard, Shames, and Teele Reference Bernhard, Shames and Teele2021; Piscopo Reference Piscopo2018).
It thus appears that the outcome—“nascent ambition”—on which we observe treatment effects constitutes an initial, key stage in the representation pipeline. Under what conditions might we expect Muslims’ nascent ambition to translate into their “expressed ambition” (“actually competing in an election,” Bernhard, Shames, and Teele Reference Bernhard, Shames and Teele2021, 380) and subsequently to them being chosen to stand and winning election? In Section E.4 of the Supplementary Material, we review theorized obstacles to marginalized group underrepresentation, as well as which are most likely to cause pipeline leakage for Muslims in India and under what conditions.Footnote 38
Based on this investigation, we speculate that regions with larger and more socio-economically privileged Muslim populations, more favorable social attitudes toward Muslims, and which feature a greater number of viable parties competing for Muslims’ votes, will be most likely to see Muslim nascent aspirants overcome posed obstacles to ultimately become elected leaders. Where Muslims constitute a greater portion of the population and individuals hold more favorable views toward the Muslim community, party gatekeepers and voters may view Muslim aspirants as more electorally competitive and desirable and may be more likely to support their candidacy (Farooqui Reference Farooqui2020; Heath, Verniers, and Kumar Reference Heath, Verniers and Kumar2015). We might similarly expect greater gatekeeper receptivity to Muslim candidates where there is more party competition over Muslim votes, as compared with a context where minority votes are largely “captured” by one party which faces little real competition (Silva and Skulley Reference Silva and Skulley2019). Importantly, if in these contexts Muslim nascent aspirants in turn anticipate less discrimination and a higher likelihood of electoral success, they may also become more likely to actually run for office. Lastly, we might expect improved representation outcomes for nascent aspirants in constituencies with more socio-economically privileged Muslims.Footnote 39 Research indicates that resource disadvantages can particularly increase the drop-off from nascent to expressed ambition amongst marginalized individuals, who are often less able and willing to make the necessary investments—and to draw on the necessary social and political capital—to successfully run for office (Bernhard, Shames, and Teele Reference Bernhard, Shames and Teele2021; Fox and Lawless Reference Fox and Lawless2010; Piscopo Reference Piscopo2018).
In sum, then, we expect that increases in Muslims’ nascent ambition caused by inclusive narratives are likely to translate, in the aggregate, to greater Muslim descriptive representation in leadership positions. Can we thus conclude that the overall consequences of inclusive narratives on Muslim descriptive representation will be positive? Consistent with this conclusion, we find scant evidence that inclusive narratives produce any countervailing effects: while they fail to move Hindu respondents, they also provoke little backlash.Footnote 40 Of course, there still may be individuals who perceive negative repercussions from inclusive narratives: for instance, aspirants encouraged by such narratives who subsequently fail to secure leadership positions may become more discouraged than they would have had they never put themselves forward. But as concerns descriptive representation, our findings indicate that inclusive narratives can lead to more underrepresented individuals in leadership positions. Given strong normative justifications for descriptive representation as well as evidence that it results in improvements in substantive representation, the value of such increases for an underrepresented community at-large should not be understated (Arnesen and Peters Reference Arnesen and Peters2018).
CONCLUSION
Across the world, activists and politicians are engaged in debates over which version of history should be enshrined in the official record. In this article, we propose that historical representations can have important implications for which groups are perceived as central to the nation, and, consequently, whether individuals from those groups feel entitled to speak on its behalf. Using real, state-sponsored Indian textbook material for our experimental treatments, we find support for our theory: Inclusive historical narratives increase Muslim participants’ perceived centrality and entitlement, their willingness to lead, and their demand for real-world Muslim leaders. These findings indicate that the deployment of history may carry important, tangible consequences and should be viewed as more than empty symbolism. At the same time, we find that Hindu participants are largely unaffected in their demand for Muslim leaders.
To what extent do our findings speak to debates over other representations of nationhood, such as statues and street names? Our theoretical and empirical focus in this paper is the effects of narratives, rather than the medium through which they are conveyed. There are many ways through which the state can present itself to its citizens and attempt to anchor a particular understanding of national identity in the public consciousness. We thus expect our findings to potentially speak not only to narratives conveyed via our specific application—textbook content—but also to the consequences of other “symbolic” or “expressive” nation-building policies such as changes to national symbols, street names and statues. While we sought to imitate textbook material to the greatest extent possible, future work could build on our findings and further extend their ecological and external validity by embedding narratives into a developed school curriculum. Although we can only speculate, we would expect that our findings would be strengthened under such a design, as students would receive greater exposure to treatment content and would have less prior exposure to alternative narratives.Footnote 41
Our findings contribute to many lines of research that, while related, are rarely in conversation, such as those on candidate emergence (Gulzar Reference Gulzar2021; Portmann and Stojanović Reference Portmann and Stojanović2019), national belonging and behavior (Abdelgadir and Fouka Reference Abdelgadir and Fouka2020; Simonsen Reference Simonsen2021), and responses to exclusionary policies (Weiss, Siegel, and Romney Reference Weiss, Siegel and Romney2022). By shedding light on the behavioral consequences of different narratives of nationhood, our findings also improve our understanding of “when, why and how national narratives of which types hinder or help democracy” (Mylonas and Tudor Reference Mylonas and Tudor2021, 115) and speak to a large literature on boundary making and its political consequences (Reicher and Hopkins Reference Reicher and Hopkins2001; Skey Reference Skey, Jackson and Molokotos-Liederman2014; Wimmer Reference Wimmer2013). More broadly, our work speaks to an important debate on the value of ethnic recognition, which has informed both philosophical (Barry Reference Barry2002; Fraser and Honneth Reference Fraser and Honneth2003; Kymlicka Reference Kymlicka1995; Taylor Reference Taylor1992) and empirical work (Cederman, Gleditsch, and Wucherpfennig Reference Cederman, Gleditsch and Wucherpfennig2017; King and Samii Reference King and Samii2020), centering largely on how ethnic federalism can improve efficacy and representation (McGarry and O’Leary Reference McGarry and O’Leary1994; Reference McGarry, O’Leary and Noel2005) and how exclusionary ethnopolitical configurations can cause political violence. For instance, Wimmer, Cederman, and Min (Reference Wimmer, Cederman and Min2009) and Cederman, Wimmer, and Min (Reference Cederman, Wimmer and Min2010) argue that ethnic nationalism promotes internal conflict and civil war by marginalizing certain ethnic groups from state power. While the authors treat variables such as changes in power structures and the political relevance of different ethnic categories as exogenously given, our research describes how the use of different narratives can alter perceptions of national identity, thereby affecting groups’ claims to the state.
While our findings represent an important contribution to the field, additional questions remain. Do our theory and findings carry across different marginalized groups and contexts? Can narratives about contemporary, as opposed to historical, contributions to the nation similarly change perceptions of belonging and entitlement to lead? Can inclusive narratives be designed such that dominant groups increase their demand for marginalized individuals as leaders? Overall, more work remains to gain a full understanding of whether and how different historical representations—transmitted through textbooks, statues, songs, or symbols—can impact the supply of and demand for marginalized individuals in leadership positions.
SUPPLEMENTARY MATERIAL
The supplementary material for this article can be found at https://doi.org/10.1017/S000305542300117X.
DATA AVAILABILITY STATEMENT
Research documentation and data that support the findings of this study are openly available at the American Political Science Review Dataverse: https://doi.org/10.7910/DVN/F8SRTA.
Acknowledgments
The authors would like to thank Sofia Breitenstein, Sabine Carey, Carl Muller-Crepon, Florian Foos, Vasiliki Fouka, Patrick Kraft, Krzysztof Krakowski, Rajeshwari Majumdar, Vittorio Merola, Salma Mousa, Eva Ostergaard-Nielsen, Bhumi Purohit, Johanna Rickne, Harald Schoen, Kristina Bakkaer Simonsen, and Chagai Weiss, as well as participants at the NYU CESS Experimental Political Science Conference, APSA, ECPR, EPSA, EuroWEPS, MWEPS, ISPP, and faculty at IE University, the University of Barcelona and Aarhus University for helpful comments, and Marten Appel for research assistance. Haas gratefully acknowledges funding support from the Centre for the Experimental-Philosophical Study of Discrimination at Aarhus University and from the Danish National Research Foundation (DNRF144). Our study was pre-registered with the Center for Open Science.
AUTHOR CONTRIBUTIONS
Authors listed in alphabetical order. Both authors contributed equally to this work.
FUNDING STATEMENT
This research was funded by Aarhus University.
CONFLICT OF INTERESTS
The authors declare no ethical issues or conflicts of interest in this research.
ETHICAL STANDARDS
The authors declare that the human subjects research in this article was reviewed and approved by Aarhus University’s Research Ethics Committee (Institutional Review Board) and certificate numbers are provided in the APSR Dataverse. The authors affirm that this article adheres to the APSA’s Principles and Guidance on Human Subject Research.
Open access
Comments
No Comments have been published for this article.