Update Article: 294

Original Title

Sanitized Title

Clean Title

Source ID

Article Id01

Article Id02

Corpus ID

Dup

Dup ID

Url

Publication Url

Download Url

Original Abstract

This study explores whether crime concerns contributed to the unexpectedly large margin of rejection in the 2022 Chilean Constitutional Plebiscite. Although the Draft Constitution aimed to address the economic concerns raised by the 2019 Social Unrest (Estallido Social), crime may have been perceived by low-income voters as a more tangible consideration. The study finds that, while crime concerns are associated with decreased support for the Draft Constitution, they do not necessarily shape the relationship between poverty and support for the Draft Constitution. In fact, I find limited evidence suggesting an already-existing negative relationship between poverty and support. Low-income voters therefore may have perceived the progressive draft as yet another example of Chilean political actors’ inability to understand the public’s priorities. However, the negative relationship between crime and support for the Draft Constitution appears to be channeled through support for President Gabriel Boric. Political cues thus seem to have played a pivotal role in the 2022 Plebiscite

Clean Abstract

Tags

Original Full Text

Crime Control or Redistribution?: How CrimeConcerns Affected Voting Behavior in the 2022Chilean Constitutional PlebisciteBy Aidan AguilarSenior ThesisDepartment of Political ScienceColumbia UniversitySpring 2024Aguilar 1AbstractThis study explores whether crime concerns contributed to the unexpectedly large margin ofrejection in the 2022 Chilean Constitutional Plebiscite. Although the Draft Constitution aimed toaddress the economic concerns raised by the 2019 Social Unrest (Estallido Social), crime mayhave been perceived by low-income voters as a more tangible consideration. The study findsthat, while crime concerns are associated with decreased support for the Draft Constitution, theydo not necessarily shape the relationship between poverty and support for the Draft Constitution.In fact, I find limited evidence suggesting an already-existing negative relationship betweenpoverty and support. Low-income voters therefore may have perceived the progressive draft asyet another example of Chilean political actors’ inability to understand the public’s priorities.However, the negative relationship between crime and support for the Draft Constitution appearsto be channeled through support for President Gabriel Boric. Political cues thus seem to haveplayed a pivotal role in the 2022 Plebiscite.I. IntroductionBeginning in October 2019, Chilean citizens’ deep discontent with the existing 1980Constitution came to the forefront of public discourse. The Constitution, drafted under thePinochet dictatorship, embraced a neoliberal model and a famously high degree of privatization.Chile’s economic model generated a high degree of economic inequality, making it the secondmost economically unequal country in the OECD (OECD 2021). These grievances brought anestimated 3.7 million total protestors, representing almost a fifth of the country’s population, tothe streets of major cities (La Tercera 2019). While initially sparked by an increase in bus fares,the protests represented long-standing discontent. The 2019 Social Unrest revolved aroundconcrete issues including the pension system, health care, education, salaries, and a newAguilar 2constitution (Ansaldi & Pardo-Vergara, 2020; Castiglioni, 2021; Somma et al., 2021; Yáñez,2021).The Draft Constitution, released in July 2022, addressed these demands through a varietyof mechanisms. Its articles included the right to housing, food, digital connectivity, and more.The Constitutional Convention proposed institutional changes to facilitate future reforms,representing a notable shift from Pinochet’s paradigm of insulating the government from“populist” impulses (Rhodes-Purdy & Rosenblatt 2023). However, despite the perceivedalignment between the long-standing grievances of protestors and the progressive draft, it wasrejected by approximately 62 percent of voters in the mandatory 2022 Plebiscite (ServEl 2022).This electoral result contrasts sharply with the demands of protesters and the results of the 2020Plebiscite, in which 78 percent of voters were in favor of drafting a new constitution (ServEl2020).This study examines whether crime played a pivotal role in preventing the sought-aftereconomic redistribution. A variety of polls conducted by the Center of Public Studies (Centro deEstudios Públicos) illustrate the increasing importance ascribed to crime by the public. Whenasked to name the three issues the government should prioritize, the share of the response“crime, assaults, robbery” rose from 26 percent in December 2019, to 42 percent in August 2021,and finally to 50 percent in May 2022 (CEP 2022). The economic complaints related to theprotests (health care, education, pensions, and salaries) were displaced by crime concerns withintwo years of the initial protests. Several scholars similarly point to the rising threat of crime inChile (Luna 2018; Sampó & Troncoso 2023; Feldmann & Luna 2023).Like low pensions and inadequate access to health care, crime may serve as a tangibleaspect of low-income voters’ lived experience. The draft may have been perceived to aggravateAguilar 3crime concerns in several manners. These potential links include the incumbent Boricadministration’s support for the draft (Berg Bilbao & Navia 2023; Hagerty 2006), theestablishment of a separate Indigenous justice system, a progressive approach to crime, rumoredfiscal irresponsibility (The Economist 2022), and the highly publicized ConstitutionalConvention (Palanza & Sotomayor Valarezo 2023). Lingering frustration surroundinginstitutions’ inability to address crime may have led low-income voters to perceive theconstitutional process as representative of the political elitism that initially sparked the protests.I use a two-level approach to explore crime’s role in the 2022 Constitutional Plebiscite. Ifirst conduct a municipal-level regression of support against crime rates, poverty, and aninteraction between the two. Higher crime rates are expected to decrease support due to thepotential links between crime concerns and the Draft Constitution. On the other hand, theMeltzer-Richard Model (1981) of inequality and redistribution would posit that municipalitieswith higher rates of income-based poverty, indicating more low-income residents, would containa higher proportion of voters in favor of the redistributive Draft Constitution. Those on thehigher end of the income spectrum may perceive economic redistribution as a burden due tohigher taxes, while those on the lower end of the spectrum would benefit from expandedeconomic benefits. Finally, given that low-income voters are disproportionately exposed to theeffects of crime, I expect crime to mediate the expected relationship between poverty and supportfor the Draft Constitution. Specifically, low-income voters may vote more similarly to theirhigh-income counterparts by rejecting the Draft Constitution as crime concerns increase. Isubsequently conduct a similar analysis using individual-level survey data obtained from theCenter of Public Studies (Centro de Estudios Públicos), and I include additional models toexplore some of the mechanisms described in the paper.Aguilar 4I ultimately find that support of President Gabriel Boric played a pivotal role in the 2022Constitutional Plebiscite. This result is consistent with the findings of Berg Bilbao & Navia(2023) and Gonzalez Ocantos & Melendez (2023). Concerns regarding increased crime werefound to have a significant and negative relationship with support for the Draft Constitution, yetthe two variables are correlated indirectly through support for President Boric. Crime concernsare associated with decreased support for President Boric, which is in turn positively associatedwith support for the Draft Constitution. Although the Draft Constitution sought to address theeconomic complaints of those impacted by the neoliberal model, this study does not find apositive relationship between poverty and support for the Draft Constitution. I in fact obtainlimited evidence pointing to decreased support for the Constitution among low-income voters.This surprising result potentially suggests a broader frustration with political elitism. I similarlydo not find strong evidence in support of the notion that crime negatively impacted the marginaleffect of poverty on support for the Draft Constitution, as the relationship was already negativein multiple specifications. I additionally find that an increase in protests, when including allcontrols, is associated with decreased support for the Draft Constitution. Relatedly, I obtainlimited evidence suggesting that an increase in protests further decreases the marginal effect ofpoverty on support for the Draft Constitution. Limitations of the municipal-level data, however,prevent this study from ascertaining whether this effect is driven by low-income votersthemselves or their high-income counterparts. Probes into the mechanisms regarding the contentof the draft itself, including those related to the establishment of a separate Indigenous justicesystem and increased social conservatism among low-income voters, do not yield conclusiveresults.Aguilar 5The paper is organized as follows. Section II establishes why discussions surrounding therejection of the Draft Constitution should incorporate the impact of crime concerns. Section IIIoutlines how poverty relates to crime concerns and voting behavior in the 2022 Plebiscite.Sections IV and V outline the parallel methodologies used to test the hypotheses and presenttheir results. Section VI tests potential mechanisms underlying the relationship between crime,poverty, and support for the Draft Constitution. Section VII discusses the implications of thecombined results. Finally, Section VIII concludes by outlining limitations and potential avenuesfor future research.II. The Growing Relevance of Crime in Chilean PoliticsExisting literature points to a wide range of factors that could have affected the success of theDraft Constitution, yet the public’s increased preoccupation with crime warrants further research.Periodic polls conducted by the Center of Public Studies ask respondents to name three issuesthat the government should address. While pensions, healthcare, education, and salaries ranked atthe top of these concerns in 2019, they were displaced by “crime, assault, and robberies” byAugust of 2021 (CEP 2022; CEP 2023). This trend continued through the ConstitutionalPlebiscite in September 2022 (CEP 2023). Given that crime outranked the other factors thatsparked the wide-scale protests in 2019, I examine whether it took precedence over economicredistribution in the 2022 Plebiscite.A variety of authors highlight the increased perception of crime in Chile. Feldmann &Luna (2023) stress that the nation faces a growing threat of violence and corruption in recentyears. This increase in crime contrasts with the long-standing perception of Chile as one of thesafest countries in Latin America (Global Peace Index 2023). Luna (2018) expands on the threatof organized crime in Chile, citing a variety of phenomena including the movement of illicitAguilar 6products in the North, an underground drug market, and the establishment of “liberated zones”where police forces struggle to operate. Sampó & Troncoso (2023) argue that this issue maydeteriorate in the future as criminal organizations exploit Chile’s vulnerabilities. For example,the nation’s geography provides thousands of miles of coastline. Given that transportation hubsfor the “transnational illicit economy” experience increased violence as groups compete forcontrol, Chile’s coastline potentially presents an opportunity for future concern (Yashar 2018).Prior studies have established that crime concerns inform vote choice. Putnam (2014),using data from Mexican presidential elections, finds the impact of perceptions of crime toinfluence vote choice up to five times as strongly as economic considerations. Hagerty (2006),focusing on the United States’ presidential elections, similarly finds that voters factorperceptions of crime into voting decisions. He specifically points to the mechanism of the“Responsibility Hypothesis,” in which voters punish or reward incumbents for their handling ofcrime.In fact, previous literature has already documented this relationship within the context ofthe 2022 Constitutional Plebiscite. Berg Bilbao & Navia (2023) find that voters used theplebiscite to express their preferences on issues not explicitly within the scope of the vote. Usingsurvey data, the authors explore the intersection of crime concern and party affiliation, and theyfind that political moderates used the plebiscite to punish the Boric administration’s handling ofcrime (Berg Bilbao & Navia 2023). This argument is consistent with that of Gonzalez Ocantos &Melendez (2023), who find that the popularity of the Boric administration served as a keysecond-dimension consideration in the 2022 Plebiscite. Given that the success of the DraftConstitution was an important factor to achieving President Boric’s policy goals (GonzalezOcantos & Melendez 2023), voters may have jumped at the opportunity to express theirAguilar 7dissatisfaction with their perception of increased crime. Voters may have believed that approvingthe Draft Constitution, handing President Boric a political victory, would fail to signal themagnitude of their discontent with the administration’s inability to curtail rising crime. Thismechanism is consistent with Hagerty (2006)’s “Responsibility Hypothesis.” In line with BergBilbao & Navia (2023), I hypothesize that increased crime concerns are associated withdecreased support for the Draft Constitution.Hypothesis 1: Crime concerns are associated with decreased support for the Draft Constitutionin the 2022 Constitutional Plebiscite.I hope to complement the study conducted by Berg Bilbao & Navia (2023) in two ways.First, this section will propose additional potential links between the Draft Constitution andcrime concerns. Second, the subsequent section will further develop the relationship betweencrime concerns and poverty. While Berg Bilbao & Navia (2023) examine the varying effect ofcrime concerns on voters of different political ideologies, this study instead focuses on itsvarying effect on voters with different incomes. Determining whether crime has a stronger effecton low-income voters may shed insight into why the redistributive Draft Constitution failed amidwidespread demands for economic redistribution.In addition to President Boric’s unpopularity, Article 309 of the Draft Constitutionsuggests another potential link to crime concerns. This article controversially established aparallel Indigenous justice system. While the article sought to acknowledge theself-determination of these communities, it received a large degree of backlash from the public.Alemán & Navia (2023) acknowledge that this vague article left room for interpretation as tohow this distinction would be applied. Given the violent clashes between militarized police andMapuche communities in La Araucanía region (Newman 2021), this clause could have increasedAguilar 8the uncertainty surrounding this violence. In fact, the establishment of this parallel systemgenerated so much debate that the Boric administration agreed to revise the Draft Constitutionpost-approval and clarify that the Indigenous justice system would not have jurisdiction overcriminal offenses (BBC 2022). This uncertainty may have motivated voters to reject the DraftConstitution.Next, the violent protests associated with the demands for a new constitution may haveweakened voters’ support for the resultant draft. Sazo (2023) argues that citizens who observedthe destruction of public spaces became increasingly concerned about the unrest. He suggeststhat this concern decreased citizens’ approval of protests as a “valid mechanism to expressdiscontent” and drove citizens to increasingly prioritize public order over civil liberties (Sazo2023). Voters may have negatively associated the draft with the social process which brought itinto existence.In addition to the Indigenous justice system, the Draft Constitution broadly embraced amore progressive approach to combating crime. Article 53, for example, compels the state toequitably enforce the law and focus primarily on the prevention of crime. Article 338 explicitlystates that the State is responsible for the reinsertion of those with prior convictions. Article 297further establishes the police as a non-military force, and it focuses on “public security”(seguridad pública) rather than the current Consitution’s “public order” (orden público). JorgeAraya argues that this distinction shifts police focus from broad surveillance towards a narrowerfight against crime (La Neta 2022). Vergara (2022) adds that the draft gave new rights to thosewho are incarcerated, including the prohibition of double jeopardy and the right to legalrepresentation. These proposed reforms occurred in response to the human rights abusesperpetrated by the Carabineros (Chile’s law enforcement) against protestors during the 2019Aguilar 9social unrest (Sazo 2023). However, they nonetheless indicate a new approach to fighting crimeand could provoke uncertainty about their efficacy. Cullen et al. (1999) highlight thatprogressives in the United States have historically struggled to effectively promote a replacementfor conservative’s “tough” approach towards those with convictions. This difficulty ofconvincing the population of a progressive anti-crime approach may play a similar role in Chile.Finally, the Draft Constitution may have been perceived to create a weaker state unable toeffectively fight crime. The Constitutional Convention itself could have contributed to concernssurrounding the strength of the draft. Castiglioni (2021) suggests that the high proportion ofindependents in the constitutional process may have threatened its legitimacy, as parties have theinstitutionalization necessary to navigate complex processes. Independents may have thusstruggled to earn voters’ confidence as they acclimated to the political stage. Luna (2018)similarly maintains that absolute transparency can disrupt the illusion that is necessary for stabledemocracies. The live streaming of the convention highlighted a series of disorderly events suchas chanting during the national anthem (Alemán & Navia 2023) and a delegate wearing aPikachu costume (The Economist 2021). Together, these occurrences may have disrupted theillusion described by Luna (2018). Palanza and Sotomayor Valarezo (2023) argue thatperceptions of mayhem surrounding the process ultimately transferred to the draft itself.Furthermore, the establishment of public services, an important redistributive element of theDraft Constitution, inherently increases the responsibility placed upon the state. The Economist(2022) labeled the draft as fiscally irresponsible, which could provoke wider concerns about thestate’s ability to fund future activities such as an effective police force. This concern, coupledwith the disorder of the Constitutional Convention, may have provoked concerns of a futureweak state. Even if scholars were to disagree with the accuracy of The Economist’s statement,Aguilar 10the messaging alone could have been crucial in the formation of voters’ opinions. Lupia (1994)argues that California referendum voters evaluating an insurance reform initiative relied heavilyon cues to make an informed vote. The length of the Chilean Draft Constitution likely forcedvoters to rely on cues in an even stronger manner. Berg Bilbao & Navia (2023) argue that these“information shortcuts” played a large role in the 2022 Plebiscite. Publications such as TheEconomist labeling the draft as fiscally irresponsible may have served as a cue strong enough todecrease support. Ultimately, a future weak state may be perceived as unable to effectively fightcrime. Luna (2018) creates an explicit link between fiscal irresponsibility and crime, arguing thatthe “relative economic weakness” of a state can stimulate the growth of criminal organizations.Aguirre & Herrera (2013), examining the Mexican state of Michoacán, similarly suggest thatinstitutional weakness can foster crime. This increased crime may in turn further financiallyweaken the government, as Kapstein & Tantravahi (2021) find that crime levels impact interestrates available to states. Crime consequently may create a cycle of state weakness.Together, these mechanisms suggest multiple possible links between the DraftConstitution and the threat of increasing crime. This study ultimately finds that President Boric’spopularity plays a pivotal role in voting decisions, consistent with the conclusion of Berg Bilbao& Navia (2023). However, I include the additional mechanisms to explore whether voters mayhave perceived the Draft Constitution itself as worsening crime, rather than only using theConstitutional Plebiscite to punish the incumbent administration.III. Poverty’s Relationship with the Constitution and CrimeIn addition to testing whether crime impacted vote choice during the 2022 ConstitutionalPlebiscite, I specifically examine how these concerns interacted with the draft’s efforts towardseconomic redistribution. Original demands for a new constitution centered around discontentAguilar 11with the neoliberal model of the 1980 Constitution. Although the 2022 Draft Constitutionaddressed these demands through a variety of mechanisms, it was nonetheless rejected by a largemargin. This section provides further context as to why this result is surprising, particularlygiven the large degree of economic inequality in Chile. Ultimately, I argue that this result may beexplained by low-income Chileans’ greater exposure to the harmful effects of crime. As a result,I expect crime to mediate the expected relationship between poverty and rejection of the DraftConstitution. I therefore anticipate that the Meltzer-Richard (1981) Model of inequality andredistribution holds only in the absence of crime concerns.It is first important to establish why the 2022 Draft Constitution would have otherwisegained the support of low-income voters. The 1980 Constitution was drafted under the Pinochetregime, whose famously neoliberal policies were influenced by the Milton-Friedman-trained“Chicago Boys” (Brender 2010). These economists advocated for the privatization of mostpreviously state-owned enterprises, most notably pensions (Borzutzky 2010). This modelbecame the global example for the limited-state models embraced by Margaret Thatcher andRonald Reagan (Barton 2014). While Chile’s model produced explosive economic growth,Borzutzky (2010) argues that their market-oriented reforms only benefited those who werealready privileged. This argument is readily visible in the nation’s economic metrics: Chile’sGini coefficient of 44.9 is the second highest of any OECD country (World Bank 2020; OECD2021).Surprisingly, Pinochet’s departure from power was not accompanied by a change in thisradical privatization. The answer to this puzzle lies in the 1980 Constitution. The Constitutionentrenched the dictatorship’s neoliberal values through a limited democracy (Ansaldi &Pardo-Vergara, 2020; Rahat & Sznajder, 1998). These “electoral engineering” practices includeAguilar 12supermajority legislative requirements and a binomial electoral system favoring the right (Rahat& Sznajder 1998). Heiss (2021) maintains that prior efforts to dismantle this system, such as theend of the binomial system, have failed to overcome its “democratic deficits,” particularly thesupermajority required to revise the Constitution’s privatized model. Because the existingConstitution is so difficult to amend, citizens increasingly recognized it as the “limiting factor”of addressing their complaints (Escudero & Olivares 2021; Fuentes 2021).This long-standing discontent serves as important context for the widespread protestsover a deceptively small hike in the bus fare. Pinochet and the Chicago Boys’ enduringneoliberal approach impeded social mobility (Somma et al. 2021). Brunner (2021) argues thatthis inequality persists despite the expansion of educational opportunities, and Morales Quiroga(2021) suggests that the government’s ability to combat inequality resulted in large-scalefrustration. A variety of studies have examined the demands of the protests themselves.Protesters largely focused on pensions, health care, education, salaries, and a new Constitution(Ansaldi & Pardo-Vergara, 2020; Castiglioni, 2021; Somma et al., 2021; Yáñez, 2021). Many ofthese issues are directly related to Chile’s privatized model. Pensions, which are distributedthrough private Administradores de Fondos de Pensiones de Chile (AFPs), struggle to provideadequate coverage for retirees (Berstein et al. 2006). Castillo-Laborde (2017) argues that Chile’smixed health insurance model favors those with ability to pay and leaves other segments withoutadequate financial protection. The government has been similarly unable to establish equitableaccess to education due to the “legal confines of an education system constructed under theneoliberal model which was introduced by the military government” (Matear 2007). Together,these tangible issues highlight the avenues through which the dictatorship’s neoliberal modelcontinues to stifle attempts to address income inequality.Aguilar 13These complaints, however, are situated in a broader frustration with the political elite’sdisregard for the middle and lower classes. Economic inequality and political exclusion thusbecome intertwined. Somma (2017) warns against attempting to separate citizens’market-oriented frustrations from broader discontent. Several authors echo this argument.Ansaldi & Pardo-Vergara (2020) maintain that the protests were not the result of a single-issuemovement, and Castiglioni (2021) points to the diverse complaints of the middle class as adriving force behind the social unrest. Silva (2021) suggests that increases in citizens’ standardof living makes it unlikely that they would mobilize on neoliberalism-related complaints alone.Fuentes (2021) offers a potential solution to this puzzle, arguing that the combination of botheconomic inequality and perception of elite abuse resulted in the large-scale protests. I ultimatelypropose that low-income voters’ broader frustration with elite abuse is largely linked to crimeconcerns.From a practical perspective, this political frustration begins with institutions themselves.Velasco & Funk (2020) point to declining trust in institutions as the catalyst for citizens’discontent. They posit that, even if the nation remains economically unequal, citizens willcontinue to comply with institutions as long as they believe the institutions work to address thisissue. Once institutions are no longer trusted to achieve their goals, however, citizens feel moreempowered to express their anger (Velasco & Funk 2020). Luna (2018) argues that thisfrustration with institutions is explicitly linked to economic inequality. Given the poor conditionsthat many low-income communities are forced to navigate, he argues that these formalinstitutions are increasingly perceived as illegitimate and absent in these communities (Luna2018).Aguilar 14Chile’s political parties have experienced a similar level of popular scrutiny. After the endof the dictatorship, the country largely split into two political coalitions: Alianza por Chile (laterrenamed Chile Vamos) and Concertación por la Democracia (Alemán & Saiegh 2007). Whilethe left-leaning Concertación governed Chile for the decades following the end of the regime,the right-leaning Alianza obtained power in 2010 through the presidency of Sebastián Piñera(Briones & Dockendorff 2015). The 2019 social unrest occurred under the leadership of Piñera,yet citizens were similarly frustrated with the leadership of the Concertación. Protests weretherefore conducted outside of parties and lamented their disconnect with the people (Heiss2021; Rhodes-Purdy & Rosenblatt 2023). Rhodes-Purdy & Rosenblatt (2023) explain thatPinochet’s regime attempted to insulate its conservative policies from the future desires of“populists,”creating a high level of political elitism. The left-leaning Concertación thus strugglesto effectively represent the electorate’s discontent. Two adjacent political phenomena highlightthis disillusionment. First, Tricot (2021) highlights the growth of territorial assemblies in Chile toillustrate the growing distrust of individuals with political parties. Second, during the first roundof the 2021 Presidential Election, the candidates of the center-right and center-left coalitionswere both outperformed by candidates from more extreme ends of the political spectrum (ServEl2021). Prior research has treated the emergence of more parties as a measure of politicalfragmentation (Laakso & Taagepera 1979), signaling a discontent with the existing parties.While this political frustration may appear as tangential to the economic complaints ofthe protests, the two issues are inseparable. Those who bore the burden of economic inequality inChile were unable to effectively express their desires through existing political channels(Rhodes-Purdy & Rosenblatt 2023). Addressing the inequality created by the 1980 ConstitutionAguilar 15requires an overhaul of the political system it enacted to protect the economic model. The 2022draft attempted to address both of these concerns through several mechanisms.First, the composition of the Constitutional Convention reflected the demands ofprotesters. The 155-member assembly established gender parity, reserved 17 seats for Indigenouscitizens, and included a range of activists (UNDP 2021). It further departed from prior politicalprocesses by consisting largely of independent delegates (Rioja et al. 2022). Belmar et al. (2023)credit this low participation of established parties due to citizens’ distrust of them. The protestshighlighted this frustration with parties, making the independent composition of the assembly animportant factor in its perceived ability to overhaul the Constitution. In addition to the identitycomposition of the assembly, its general attitude reflected that of the protests. These delegatesembraced what Verdugo & Prieto (2021) refer to as “aversive constitutionalism,” as theycritically evaluated how the 1980 Constitution created harm to avoid repeating those mistakes inthe future. The assembly itself therefore initially appeared to embody the demands of the 2019mass mobilization.Once these deliberations were completed, the Draft Constitution itself included directresponses to protesters’ neoliberalism-related complaints. The BBC (2022) provides an effectivesummary of the economic implications of the 162-page document. The existing Constitutionactively restrained the state from participating in market activities, while the 2022 DraftConstitution established Chile as a “Social and Democratic State of Law.” The draft alsostipulates that the State must satisfy the basic needs of citizens, and it establishes public healthand social security systems (BBC 2022). Its articles include the right to housing (Art. 51), food(Art. 56), digital connectivity (Art. 86), a clean environment (Art. 104), and more. These explicitrights represent a shift from the previously privatized model. For example, Article 57Aguilar 16(establishing the right to water and sanitation) sharply contrasts with the nation’s privatizedwater system. Baer (2014) argues that, although Chile narrowly fulfills the human right of accessto water, it only does so because of heavy state involvement in the privatized resource. Clearlyoutlining this right in the Draft Constitution bolsters the state’s ability to limit the power ofprivate companies in administering this human right.In addition to these responses to demands explicitly related to economic inequality, theDraft Constitution included structural changes to facilitate future demands and addresscomplaints regarding political elitism. These changes included the replacement of the Senate bya less powerful “Chamber of Regions” to remedy the limited democracy that protected theextreme neoliberal model (Reuters 2022). Together, these changes suggest that both theeconomic and political grievances of protestors were heavily prioritized when drafting the 2022Draft Constitution.The subsequent 62-percent rejection of the draft is therefore surprising for two reasons.First, it appears initially inconsistent with the demands of the public. Chile’s police forceestimates that almost a fifth of the country’s population demonstrated in the streets (La Tercera2019). Castillo et al. (2015) maintain that an individual’s beliefs about economic distribution arean important factor in determining their participation in protests. The Draft Constitution centeredthese economic complaints when drafting the articles outlined above. Furthermore, 78 percent ofvoters in the 2020 Plebiscite supported drafting a new constitution (ServEl 2020). One couldpossibly attribute this difference in support to the plebiscites’ different participation rates (51percent versus 86 percent of Chile’s population), as the 2022 Plebiscite was mandatory. Thisview would suggest that many voters in the first plebiscite participated because they sought toadopt a new constitution, while the second plebiscite more accurately captured public sentimentAguilar 17by including more voters. However, a poll conducted by Espacio Público after the 2022Plebiscite suggests that this may not be the case (cited in Fuller 2023). Not only did 74 percent ofrespondents argue that the country still needs a new constitution, 57 percent of respondents whovoted to reject the Draft Constitution still believed that a new constitution was necessary(Espacio Público 2022). This sizable support among those who voted against the DraftConstitution in the 2022 Plebiscite highlights the broader electorate’s desire for a newconstitution.Second, the median voter in Chile would have presumably benefited from theredistributive elements of the Draft Constitution. Chile’s high Gini coefficient suggests thatwealth is concentrated in a small portion of the population (World Bank 2020). The majority ofthe population bears the harmful impacts of the privatized systems mentioned by Somma et al.(2021). The Meltzer-Richard Model (1981) posits that a welfare-maximizing voter will chooseredistribution in an economically unequal society. This median voter is in favor of redistributionwhen the mean income is greater than that of the median voter, suggesting a right-skewedincome distribution. Given that the Draft Constitution represents a nationwide initiative, Iassume that voters use a national, rather than local, notion of economic inequality. This broaderoutlook is especially relevant considering the national campaigns conducted through both“traditional and social media” (Vergara 2022). Voters may therefore compare their incomes to anational mean and median. A municipality having more low-income residents therefore implies agreater number of voters who consider themselves as earning below the national median income.Because the Meltzer-Richard Model assumes that all voters who earn less than the median voterwill similarly vote in favor of redistribution, I assume that municipalities with larger proportionsof residents living in income-based poverty have a larger share of residents who would vote inAguilar 18favor of the redistributive Draft Constitution. The mandatory nature of the 2022 Plebiscite shouldeven more accurately represent the entire continuum of voter income used in the Meltzer-RichardModel. Corvalan & Cox (2013) find that the decreasing turnout of the Chilean electorate isincreasingly class-biased in favor of those with higher incomes. Mandatory participation should,in theory, address the class disparity associated with shrinking turnout. However, the politicalreality of the 2022 Plebiscite contradicts the predictions of this model. Figure 1 suggests that therelationship between poverty and rejection may even be the opposite to what the Meltzer-RichardModel predicts. Municipalities with higher rates of rejection in the 2022 Plebiscite appear to bethose with the highest rates of income-based poverty. I propose that crime concerns are crucial inaddressing this puzzle.Figure 1: Comparing the Distribution of Poverty and Rejection at the Municipal LevelPolitical economy literature points to second-dimension considerations as obstacles to thereal-life application of the Meltzer-Richard Model. Iversen & Goplerud (2018) argue that “aAguilar 19salient noneconomic dimension” can divide low-income voters and inhibit redistribution. Priorstudies have explored other contexts and found factors such as religion (Jasiewicz 2009; De La O& Rodden 2008) and social identity (Han 2016) to take priority over economic considerations.Of course, low-income voters are not one-dimensional. Income intersects with a wide range ofother identities, and the length of the draft provides opportunities for other salient considerations.These considerations include the right to the voluntary termination of a pregnancy (Art. 61), thedefinition of Chile as a plurinational and intercultural state (Art. 1), and the establishment of aparallel Indigenous justice system (Art. 309). In the wake of the draft’s rejection, several studieshave attempted to explain the unexpectedly large margin of defeat. Both Disi Pavlic (2024) andPiscopo & Siavelis (2023) find that Indigenous communities, who received increased recognitionin the Draft Constitution and reserved seats in the assembly, rejected the draft due to feelings ofdisconnect from their representatives. Sazo (2023) offers a parallel explanation of disconnect,arguing that citizens disapproved of delegate scandals and the prioritization of certain measures.Finally, Gonzalez Ocantos & Melendez (2023) argue that the popularity of the incumbent Boricadministration, who supported the Draft Constitution, served as a key second-dimensionconsideration for Chilean voters. The unique setting of a constitutional plebiscite, especiallyconsidering the length of the draft, increases opportunities for second-dimension considerationsin voting decisions. Gillette (1988) further highlights that plebiscites shed greater insight intovoter preferences than representative politics. The 2022 Constitutional Plebiscite thereforeprovides scholars with a valuable occasion to explore the priorities of low-income Chileans.Crime’s role as a primary public concern ahead of the 2022 Plebiscite suggests that itlikely served as a key second-dimension consideration (CEP 2022). The previous sectionsuggests that the public could have associated the Draft Constitution with crime concerns acrossAguilar 20multiple dimensions. While my analysis controls for the second-dimension considerationsestablished by other studies, the salience of crime suggests that it may have been the primaryfactor that prevented the redistribution desired by the public. I thus expect the Meltzer-RichardModel to hold only in the absence of crime concerns.Hypothesis 2: When controlling for crime concerns, I expect to observe a positive relationshipbetween poverty and support, consistent with the predictions of the Meltzer-Richard Model.I propose that crime concerns not only served as a key second-dimension consideration inthe 2022 Plebiscite but that they are inextricably linked to poverty. Low-income citizens’complaints related to neoliberalism, including institutional neglect, may similarly apply to crimeconcerns. In this way, the redistribution-oriented and unprecedentedly diverse ConstitutionalAssembly struggled to overcome the political elitism that historically plagues Chilean politics.Prior literature establishes that crime, like poverty, impacts individuals’ quality of life.Hagerty (2006), for example, argues that individuals incorporate crime rates into calculation oftheir quality of life. Changes in this quality of life are used as the basis for his “ResponsibilityHypothesis” in which voters punish or reward incumbents. Hanslmaier (2013) similarly findsthat fear of crime victimization reduces the life satisfaction of survey respondents in Germany.Naturally, citizens who fear for the safety of their bodies and belongings amidst rising violencemay carry more stress than those who live in non-violent situations.Several studies have established links between economic inequality, poverty, and crime.Krahn et al. (1986) argue that the “relative deprivation” theory results in a positive relationshipbetween income inequality and homicide rates. The authors cite Zehr (1976), who finds that theaggression created by relative deprivation is largely targeted towards peers. This finding impliesthat low-income individuals suffer abuse at the hands of both the state and their disillusionedAguilar 21neighbors. Dong et al. (2020), on the other hand, contend that “absolute deprivation” resultingfrom poverty itself, rather than economic inequality compared to one’s peers, increases homiciderates. Together, both theories highlight direct links between poverty and crime. Chile’s economicmodel has created acute poverty among its lower classes as a result of economic inequality. Assuch, both “absolute” and “relative” deprivation theories may be relevant. This study treatspoverty as the extent to which an individual bears the impact of economic inequality. Chile’seconomic model both socially and geographically separates socioeconomic classes (Luna 2018),resulting in crime having a disproportionate impact on certain low-income communities. Luna(2018) maintains that crime itself is a direct result of the poverty in which a large percentage ofChile’s citizens live. Illicit activities may provide low-income individuals with an opportunity forsocial mobility amid a lack of employment opportunities (Luna 2018; Feldmann & Luna 2023).On a social level, Feldmann & Luna (2023) point out that Chilean narco culture portrays crimeas a form of rebellion against Chile’s abusive economic and political elites. Individuals frustratedwith their economic situation may therefore have yet another incentive to engage in criminalactivity.As a consequence of the link between poverty and illegal activity, low-income votersdisproportionately bear the burden of rising crime. This burden spans several dimensions. First,low-income individuals experience incarceration at a much higher rate than their high-incomecounterparts. Wilenmann (2020) offers an explanation for this phenomenon: rising incarcerationrates in Chile represent a “project of control of a large, marginalized population throughconfinement.” While this approach is certainly not limited to Chile, as Lofstrom & Raphael(2016) document the link between incarceration and socioeconomic inequality in the UnitedStates, Chile’s notable economic inequality may exacerbate this issue. Second, low-incomeAguilar 22communities bear the mental stress of their loved ones’ involvement in criminal activities. Luna(2018) highlights that low-income residents on the outskirts of Santiago, the nation’s capital, feartheir children joining the drug trade. Third, police brutality resulting from increased crime maysimilarly place a disproportionate impact on those living in high-crime areas. Bonner (2013)argues that, while the Carabineros are respected, this respect comes from fear and suppressedcriticism of their behavior. Low-income communities who experience greater exposure to crimemay similarly endure the harsh response of the Carabineros. Fourth, rising crime rates mayprevent future opportunities for economic growth. Lacoe et al. (2018) find that rising crime rateson certain city blocks results in decreased private investment. Although this study was conductedin the United States, it may still be relevant in Chile. Crime therefore inhibits the potential offuture economic growth and entrenches the cycle of violence that low-income communitiesexperience. These mechanisms illustrate the numerous ways in which low-income voters may bedisproportionately impacted by crime.The 2019 Social Unrest provides a useful case study to examine the link between thesemechanisms and the desire for a new constitution. Somma et al. (2021) highlight thatlow-income communities bore the brunt of disruption from the protests. From an economicperspective, the authors argue that looting resulted in a decrease in the availability offundamental goods in low-income neighborhoods. Looting highlights the lingering economichardships that low-income communities may face as a result of crime. Not only were theseindividuals unable to obtain the basic commodities necessary for survival, but they may nowstruggle to do so in the future. Businesses likely perceive these criminal practices as anadditional risk associated with investment in low-income communities (Lacoe et al. 2018),resulting in a future scarcity of important goods and services. This looting also triggered anotherAguilar 23mechanism of the disproportionate burden of crime on low-income communities: police brutality(Somma et al. 2021). Amnesty International (2020) argued that the Carabineros defiedinternational law in their use of excessive force against protesters. In this way, low-incomecommunities suffered the abuse of both criminals and the state. This case study helps illustratewhy low-income voters may be particularly motivated by crime concerns in voting decisions.But would these concerns have trumped the strong desire for economic redistribution thatprompted the process of drafting a new constitution? Studies suggest that crime may play a largerrole than initially predicted. As suggested by the several links between poverty and crime,low-income communities are often familiar with the tangible harms of crime. Highly tangiblefactors in the daily lives of voters have been shown to determine perceptions of their well-beingand in turn influence voting decisions. Somma (2017) argues that frustration with elites onlyresults in mobilization if citizens can articulate their discontent in specific mechanisms.Practically, these mechanisms may include certain political actors or abusive policies. Peña(2021) suggests that the importance of these tangible factors may shed insight into why theSocial Unrest erupted. Interestingly, both poverty and economic inequality, as measured by theGini coefficient, had decreased in Chile since 1990 (Hadzi-Vaskov & Ricci 2021; Peña 2021).Nonetheless, a large portion of the population mobilized against economic inequality. The key tothis contrast lies in the fact that day-to-day life retains “signs of discomfort” for most Chileans(Peña 2021). These tangible factors include the current models of pensions, healthcare, andeducation that played a central role in the protests. Crime likely serves as a similarly tangiblefactor in the voting decisions of low-income communities. As demonstrated by both Hagerty(2006) and Hanslmaier (2013), fear of crime impacts an individual’s assessment of their qualityof life. In fact, crime concerns may have even sparked the discontent that brought about theAguilar 24protests. Waissbluth (2020) suggests that jails with higher rates of recidivism and frustration withthe state’s inability to effectively address crime contributed to the unrest. Specifically, protestersfelt that these issues were not perceived by elites and were therefore not prioritized (Waissbluth2020). This disconnect relates to the segregation created by Chile’s neoliberal model (Luna 2018;Waissbluth 2020). The issue of crime was thus regarded as a tangible result of economicinequality. Given that tangible concerns motivated the initial protests, the tangible concern ofincreasing crime may have driven low-income voters to reject the Draft Constitution.Crime concerns, in addition to triggering the mechanisms discussed above, may alsocontribute to growing pessimism about the future. Sazo (2023) argues that this pessimism was adecisive force in the 2022 Plebiscite. This pessimism was linked specifically to the unrest’sviolent protests, the rise of criminal violence, and clashes with Indigenous armed groups (Sazo2023). Pessimism surrounding the creation of a document as fundamental as a nation’sconstitution likely serves as a significant roadblock to its adoption. For low-income citizens, thispessimism may have multiple effects. First, it may generate concerns about the futuregovernment’s capacity to effectively address the crime that disproportionately impactslow-income communities. Second, this pessimism may plant doubts surrounding the feasibilityof economic redistribution. The higher interest rates mentioned by Kapstein & Tantravahi (2021)offered to high-crime states may prevent them from obtaining the financing necessary to fulfillthe responsibilities outlined in the Draft Constitution. This second mechanism is a particularlycrucial factor in explaining why the tangible impacts of crime would outweigh those of economicinequality. Even if voters were to choose to prioritize economic redistribution in the present, thenegative effects of crime could inhibit the future state from fulfilling this commitment.Aguilar 25Low-income voters would be exposed to the harmful effects of both crime and the failure ofeconomic redistribution. They may therefore be more likely to reject the Draft Constitution.Hypothesis 3: I expect the effect of crime to decrease the anticipated positive marginal effect ofpoverty on support for the Draft Constitution.Previous literature regarding the intersection of crime and economic redistribution doesnot appear consistent with the results of the 2022 Plebiscite. Rueda & Stegmueller (2016) shedinsight into the link between crime concerns and economic redistribution. Focusing on WesternEurope, they find that high-income voters are more supportive of redistribution in more unequalareas due to income inequality’s “negative externality” of higher crime. The authors suggest thatlow-income voters are primarily motivated by “tax and transfer consequences” and are thereforemore consistently supportive of redistribution. The large rejection of the Draft Constitution,however, suggests otherwise. Those who would have benefitted from the draft’s “tax and transferconsequences” compose the majority of Chile’s population given its high Gini coefficient, yet thelarge margin of rejection suggests that they were not invariably supportive of redistribution. Thiscontrast calls for additional exploration into how low-income voters weigh crime concernsagainst the promise of economic redistribution.The Constitutional Convention’s inability to effectively address crime concerns may havebeen an extension of the political elitism that first inspired the protests. Political elites unable torelate to the disproportionate impact of crime upon low-income communities are less likely toadequately address it. After the rejection of the plebiscite, Sazo (2023) points out that PresidentGabriel Boric implemented changes in his cabinet to prioritize security concerns. Theseadministrative changes, coupled with the growing salience of crime in public opinion surveys,Aguilar 26suggest that crime concerns may be a crucial factor to consider in the citizens’ attempt to addresseconomic inequality.IV. Methodology 1: Municipal-Level DatasetA municipal-level analysis serves as the primary methodology for this study. This section firstdescribes the data and models used to test the hypotheses, and it subsequently presents theresults.A. DataI first test the hypotheses using a municipal-level analysis of Chile’s 346 communes. Thesemunicipalities represent the smallest unit of analysis using government-sponsored data. By usingthis government-calculated data, I hope to increase the reliability of the regressions. Thevariables were largely obtained from different sources. All data points were standardized for thepurpose of interpretability.Conceptually, the dependent variable is a municipality’s support for the DraftConstitution. This dataset measures the conceptual dependent variable through the percentage ofa municipality’s residents that voted to approve the Draft Constitution in the 2022 ConstitutionalPlebiscite. This data is obtained from the Chilean Electoral Service (ServEl), which publishesinformation on the outcomes of past elections. Proportions of approval may indicate the extent towhich the voting decisions of individuals were impacted by the relevant independent variables.The first independent variable, crime concern, is proxied by the three-year change inpolice reports filed per municipality. This data is obtained from the Center of Study and Analysisand Crime (Centro de Estudio y Análisis de Delito). For the purposes of comparison, all policereports were converted to a standardized rate of police reports per one hundred thousandresidents. Crimes were additionally filtered to include only those most visible to the broaderAguilar 27community (e.g., removing domestic violence and cattle rustling). A full list of excluded andincluded crimes is located in the Appendix.1 I focus on three-year change in crime due to thetimeline of the constitutional process. The increasing importance ascribed to crime in publicsurveys indicated shifting priorities of the electorate between 2019 and 2022. I additionallyassume that perceived deteriorations in crime would result in stronger crime concerns, as thosewho live in consistently high-crime areas may become desensitized (Di Tella et al. 2019). Priorstudies have acknowledged that crime rates (Skogan 1986) and victimization (Berg Bilbao &Navia 2023) do not necessarily equate to crime concerns, yet this data source likely serves as themost viable proxy available. Figure 2 presents the standardized distribution of three-year changesin police reports per one hundred thousand residents.Figure 2: Standardized Distribution of Three-Year Change in CrimeNote: Two municipalities, Timaukel and Perquenco, were removed as outliers. No crime datawas available for one municipality, Antártica.1 I exclude two communes (Timaukel and Perquenco) in the Magallanes y Antártica Chilena region due to theirsmall population sizes. In these communes, single-digit changes in the number of police reports were magnified to amassive change per one hundred thousand residents. Furthermore, no data was available for the Antártica commune.Because crimes were manually filtered, the population-weighted rates were also manually calculated using data fromthe 2017 Census.Aguilar 28The second independent variable, Poverty Measure, is conceptually the degree to which amunicipality is low-income. This conceptual variable is measured by the proportion of amunicipality’s residents living in income-based poverty. Given that the government does notpublish explicit data on intra-municipal inequality or residents’ incomes, I use a measure ofincome-based poverty jointly published in 2020 by the Chilean Ministry of Social Developmentand Family (Ministerio de Desarrollo Social y Familia) and the United Nations EconomicCommission for Latin America and the Caribbean. This data is obtained through thegovernment-sponsored Casen survey (Encuesta de Caracterización Socioeconómica Nacional).The Ministry of Social Development and Family uses the Fay-Herriot (1979) model to estimate acommune-level disaggregation of the data (Ministerio de Desarrollo Social y Familia 2020). Onan aggregate scale, I assume this higher incidence of poverty implies a greater number of votersbelow the national median income. Hypothesis 2 posits that a higher incidence of poverty will beassociated with a larger percentage of residents in favor of the Draft Constitution. Figure 3presents the standardized distribution of a municipality’s percentage of residents living inincome-based poverty.Figure 3: Standardized Distribution of Proportion of Residents Living in PovertyAguilar 29The third independent variable is an interaction variable between the three-year change inpolice reports and the percentage of a municipality’s residents living in income-based poverty.As discussed in the theory section, I expect crime concerns to have a negative impact on therelationship between poverty and support, as low-income individuals may be disproportionatelyexposed to the harmful impacts of crime. This disproportionate exposure may outweigh thebenefits posed by future economic redistribution.B. Model SpecificationsThe baseline model specification simply includes the three independent variables, as botheconomic and crime-related complaints appeared to play a central role in the 2022 Plebiscite:[Support for Constitution]i = β0 + β1[Crime Concerni] + β2[Poverty Measurei] + β3[CrimeConcerni x Poverty Measurei] + εi (1)Given the existence of second-dimension considerations, I include a range of controlvariables to attempt to isolate the effects of income and crime in the two regressions. I split thecontrol variables into three tiers. I first include controls in specification (2) for the other salientelements of the Draft Constitution itself, which I refer to as “intrinsic controls.” Because therewere external factors contributing to the success of the draft, I subsequently add region fixedeffects in specification (3) and finally include a control for voters’ support of President Boric inspecification (4).Beginning with intrinsic controls, I first include a control for Indigeneity proxied by thepercentage of municipality’s residents who identify as Indigenous. The Draft Constitutionofficially recognized 11 Indigenous groups and provided them with expanded rights (e.g.,parallel justice system). I therefore include this control to account for the effect of theseexpanded rights on voters’ approval or rejection of the plebiscite. This data is made availableAguilar 30through the 2017 Census. Second, I include a control to account for the reproductive rightsincluded in the Constitution. Article 61 of the Draft Constitution included sexual andreproductive rights, including the right to a “voluntary termination of a pregnancy.” This articlebecame a highly divisive issue ahead of the Plebiscite. Conservative-led misinformation includedrumors that the Draft Constitution allowed “on-demand late-term abortions” (Vergara 2022).Because low-income voters may also have been motivated by their beliefs on abortion to rejectthe Draft Constitution, I include Abortion Opinion as a control variable to isolate the relationshipbetween crime, poverty, and support. I use religiosity as a proxy for this conceptual control, asopposition to reproductive rights in Latin America has traditionally been led by religious groups(Biroli 2020). I therefore obtain the percentage of a municipality’s residents identifying asreligious from the National Congressional Library of Chile, which uses data from the 2002National Census.2I then include region fixed effects in specification (3) to control for region-specificheterogeneity. Finally, I control for the approval of President Gabriel Boric (Boric Support) inspecification (4), which proved a key second-dimension consideration in the studies conductedby Gonzalez Ocantos & Melendez (2023) and Berg Bilbao & Navia (2023). President Boric’sleft-wing agenda embodies many of the priorities held by the Draft Constitution. The reformslaid out in the Draft Constitution, including the structural changes to the government, werenecessary to achieving much of the Boric administration’s goals (Gonzalez Ocantos & Melendez2023). To proxy for approval of Gabriel Boric in the municipal-level “all controls” model, I usea commune’s percentage of voters in favor of President Gabriel Boric in the first round of the2 Although this data is twenty years old and evangelism has attracted a growing number of Catholic converts, I donot expect the level of religiosity in each commune to change drastically during that time.Aguilar 312021 Presidential Election.3 Although this variable does not explicitly capture voters’ satisfactionwith his post-inauguration performance, it nevertheless gives insight into the role played bypolitical beliefs external to the content of the draft. I also do not expect voter support forPresident Boric to change drastically within the first six months of his presidency. To mitigatethis concern, I include a more explicit question regarding Boric approval in the individual-levelmethodology described in the following section.C. ResultsThe municipal-level regressions yield mixed evidence with respect to the outlined hypotheses.While increased crime is associated with a decreased support for the Draft Constitution in mostspecifications, this effect disappears when controlling for support of President Boric. The resultsare outlined in Table 1.Table 1: Municipal-Level Regression ResultsSupport for the Draft Constitution(1) (2) (3) (4)Crime Concern -0.238*** -0.196*** -0.119*** -0.009(0.045) (0.044) (0.038) (0.020)Poverty Measure -0.470*** -0.385*** -0.013 0.022(0.046) (0.053) (0.063) (0.032)Crime-Poverty Interaction -0.024 -0.024 -0.046 -0.028*(0.036) (0.035) (0.029) (0.015)Indigeneity -0.069 0.004 0.009(0.051) (0.065) (0.033)3 There was a second round in the 2021 Presidential Election, yet this first-round vote may serve as a more effectivecontrol to capture both political leaning and opinions on President Boric himself. Because José Antonio Kast, theright-wing opponent, implied praise for Pinochet (Reuters 2021), President Boric’s support in the second round mayinclude voters who simply sought to avoid any reminder of the dictatorship.Aguilar 32Abortion Opinion -0.275*** -0.244*** -0.016(0.047) (0.045) (0.024)Boric Support 0.883***(0.029)Constant 0.015 0.007 -0.993*** 0.121(0.045) (0.043) (0.291) (0.152)N 343 338 338 338Adjusted R2 0.310 0.376 0.621 0.903Note: *p < 0.1 **p < 0.05 ***p < 0.01. All variables are standardized.The coefficient of the three-year change in crime is significant and negative in the firstthree specifications, providing limited evidence in favor of Hypothesis 1. In Column 3, a onestandard deviation increase in Crime Concern is associated with a -0.119 standard deviationdecrease in support for the Draft Constitution. For comparison, a one standard deviation increasein Abortion Opinion is associated with a -0.244 standard deviation decrease in support. Again,however, the significance of this coefficient disappears when controlling for a municipality’ssupport of President Gabriel Boric. With respect to Hypothesis 2, the negative and significantcoefficient of Poverty Measure provides limited evidence in contrast to the theory outlinedabove. The Meltzer-Richard Model predicts that a greater proportion of residents living inincome-based poverty would increase the proportion of residents in favor of the redistributiveDraft Constitution, as there would be more voters below the national median income. However,the negative coefficient of Poverty Measure suggests the opposite. Municipalities with a higherincidence of poverty generally had a smaller proportion of voters who approved the DraftConstitution. While this effect disappears after adding all controls, the evidence generally doesnot support Hypothesis 2.Aguilar 33Hypothesis 3, which posits that crime mediates the effect of poverty, receives limitedsupport in Column 4 of Table 1. This coefficient suggests that increased crime may strengthenthe negative relationship between poverty and support for the Draft Constitution. However, asillustrated by Figure 4 below, this effect is not particularly pronounced when considering themargin of error.Figure 4: Marginal Effect of Income on Draft Constitution Approval by Crime LevelNote: Values used in this graph correspond to all controls included in specification (4). Thesecontrols include standardized Abortion Opinion, Indigeneity, Boric Support, and region fixedeffects. Fewer observations with high z-scores lead to greater margins of error.While the data does not provide resounding evidence in favor of any of the hypotheses, itindicates that support of Boric played a pivotal role in the 2022 Plebiscite. This finding isconsistent with the studies conducted by Gonzalez Ocantos & Melendez (2023) and Berg Bilbao& Navia (2023).V. Methodology II: Individual-Level DataAguilar 34Recognizing the limitations of aggregate municipal data, I conduct a similar individual-levelanalysis. One could possibly critique the municipal-level relationships described in the previoussection as struggling to capture the thought process of individual voters, and this section aims tomitigate this concern. Because the municipal-level analysis already describes the fourspecifications used to test the hypotheses, this section simply describes the data used in theindividual-level analysis and subsequently presents results.A. DataI obtain survey data from the Center of Public Studies.4 The measures in this dataset are largelymore explicit than the proxies used in the municipal-level dataset. Although the original surveyincluded a total of 1441 respondents, the pool was narrowed to the 516 respondents who hadanswered questions related to the variables of interest.5 The individual-level methodology willuse a logistic regression.The dependent variable in this case will include both a respondent’s actual or intendedvote in the 2022 Plebiscite. Because this study seeks to examine the factors related to theformation of a vote choice, I include both groups as to not exclude those who were similarlyinfluenced by the independent variables of interest (and may have failed to participate in themandatory plebiscite for an unrelated reason).The first independent variable, Crime Concern, explicitly captures whether a respondentis concerned with crime. Crime concern is a dummy variable equal to 1 if the respondent listscrime, assault, robbery, drugs, and/or violence as one of the top 3 concerns that the governmentshould address, and it is equal to 0 otherwise.5 Additional survey questions were used to narrow down respondents but were not included in the final analysis. Forfuture analysis, the responses to these additional questions are included in the individual-level replication file.4 The survey (No. 88) was conducted in November and December 2022, representing a measure slightly posterior tothe plebiscite. However, the prior survey was conducted in April and May of 2022 (before the draft was officiallypublished in July). As such, the slightly posterior survey may more accurately reflect the formation of preferences asa result of the Draft Constitution itself and the surrounding campaigns.Aguilar 35The second independent variable, Poverty Measure, aims to capture the extent to which arespondent could gain from economic redistribution. Individual-level poverty, measured by theinverted scale of a respondent’s monthly income6, implies a lower position on the incomespectrum. This would increase their likelihood of being below the national median income and,according to the Meltzer-Richard Model, voting for economic redistribution. Inverting theincome scale, while potentially appearing counterintuitive, allows the individual-level regressionresults to be more easily interpreted alongside those from the municipal-level regression.Finally, the third independent variable will be an interaction variable between the twovariables. Consistent with the hypotheses, I expect crime to decrease the marginal effect ofpoverty on support for the Draft Constitution.Control variables are also more explicitly specified than the municipal-level dataset.Beginning with “intrinsic controls” in specification (2), Indigeneity is now a dummy variableequal to 1 if the respondent indicates that they pertain to an Indigenous group. I also control forAbortion Opinion, which is now a dummy variable equal to 1 if respondents answered thatabortion should only be allowed in certain cases or that it should be prohibited. Because theDraft Constitution included the right to the voluntary interruption of a pregnancy (“unainterrupción voluntaria del embarazo”), I considered both the restriction and prohibition ofvoluntary abortions as factors that may decrease voter support. I subsequently add region fixedeffects in specification (3) to control for region-specific heterogeneity. In specification (4), Icontrol for a respondent’s opinion of President Boric. I include a dummy variable equal to 1 ifthe respondent indicates an approval of President Boric, and 0 otherwise. This measure more6 The scale represents a general income measure asked across separate questions given to respondents based on thereported number of individuals in their household.Aguilar 36explicitly captures voters’ approval of the President’s post-inauguration performance. Across allspecifications, I include controls for a respondent’s age and sex.B. ResultsThe results of the individual-level regression are presented below in Table 2. The results arelargely consistent with those of the municipal-level regression, and they again emphasize theimportance of Boric support in determining voters’ choice in the 2022 Plebiscite.Table 2: Individual-Level Regression ResultsSupport for the Draft Constitution(1) (2) (3) (4)Crime Concern -2.931*** -2.598*** -2.522*** -1.207(0.874) (0.954) (0.969) (1.482)Poverty Measure -0.371** -0.270* -0.262 -0.183(0.145) (0.159) (0.164) (0.248)Crime-Poverty Interaction 0.399** 0.337* 0.322* 0.156(0.165) (0.180) (0.182) (0.272)Indigeneity 0.131 0.243 0.375(0.332) (0.343) (0.456)Abortion Opinion -1.813*** -1.825*** -1.365***(0.212) (0.223) (0.301)Boric Support 4.361***(0.393)Age -0.008 -0.001 -0.001 -0.025***(0.006) (0.006) (0.006) (0.010)Sex 0.138 -0.057 -0.017 -0.062Aguilar 37(0.187) (0.205) (0.212) (0.294)Constant 2.484*** 3.088*** 3.159** 2.602(0.844) (0.926) (1.237) (1.733)N 516 516 516 516Log Likelihood -339.527 -298.967 -293.058 -174.495Note: *p < 0.1 **p < 0.05 ***p < 0.01. Reverting the poverty scale measure to its originalincreasing income measure causes the coefficient of Crime Concern to lose its significance. Themagnitudes of the coefficients are not readily interpretable due to the poverty measure’s use of aseven-point scale rather than a binary variable.Similar to the municipal-level regression, a respondent citing a concern with crimedecreases their likelihood to vote in favor of the Draft Constitution in Columns 1, 2, and 3.However, adding a respondent’s support for President Boric causes the coefficient of crime tolose its significance. Therefore, the results again provide only limited evidence in favor ofHypothesis 1. The coefficient of the poverty measure is significant and negative in Columns 1and 2, yet its significance disappears when adding region fixed effects and controlling forsupport of President Boric. In contrast with the municipal-level regression, the coefficient of theinteraction variable between crime and poverty is significant and positive in Columns 1, 2, and 3.The direction of this coefficient does not support Hypothesis 3, and its significance disappearswhen adding all controls. I run a similar regression using a respondent’s inverted education levelas a proxy for poverty, and I find generally similar results. These coefficients are included in theAppendix and presented in Table A3.VI. Additional MechanismsThis section seeks to shed additional insight into the mechanisms underlying the results of themunicipal- and individual-level regressions. Mechanisms tested include using support forPresident Boric as an intermediate endogenous variable, the interaction between Indigeneity andAguilar 38crime concern, and the interaction between poverty and abortion opinion. Overall, the resultsprimarily suggest that crime concerns indirectly affected the 2022 Constitutional Plebiscitethrough support for President Boric.A. Boric Support as a Mediator for Crime ConcernsGiven that the significance of crime disappears only after controlling for Boric support, Isubsequently determine whether voters’ support for President Boric serves as a mechanismthrough which crime indirectly affects support for the Draft Constitution. This mechanism isconsistent with Hagerty (2006)’s “Responsibility Hypothesis” and Berg Bilbao & Navia (2023)’sanalysis of the 2022 Plebiscite. Voters’ discontent with the state of crime in Chile coulddetermine their approval of President Boric. His support for the Draft Constitution may haveprovided voters with an opportunity to express their discontent with his performance on crime.Alternatively, the length of the Draft Constitution forces voters to rely on cues (Berg Bilbao &Navia 2023; Lupia 1994). A politician whom voters dislike could serve as a sufficient signal toinform their vote. Boric support could therefore be endogenous to crime concerns.To explore whether crime concerns indirectly affected the 2022 Plebiscite throughsupport for President Boric, I conduct a mediation analysis using both the municipal- andindividual-level data. The analysis uses the Baron-Kenny (1986) approach, which defines amediator as “the generative mechanism through which the focal independent variable is able toinfluence the dependent variable of interest.” Although the phrasing of a moderator variableimplies a certain degree of causality, this study instead focuses on correlations between variablesdue to the use of observational data. Baron & Kenny (1986) prescribe three steps: regressing themediator (support for President Boric) on the independent variable (crime), regressing thedependent variable (support for the Draft Constitution) against crime, and finally regressingAguilar 39support for the Draft Constitution on both crime and support for President Boric. Given that thecoefficient of Poverty Measure loses significance when adding region fixed effects in both themunicipal- and individual-level regressions, I include it simply as a control variable in themediation analysis to isolate the relationship between crime, support for President Boric, andsupport for the Draft Constitution. Tables 3 and 4 present results.Table 3: Municipal-Level Mediation Analysis ResultsBoricSupportSupport for DraftConstitution(1) (2) (3)Crime Concern -0.123*** -0.116*** -0.007(0.037) (0.038) (0.020)Poverty Measure -0.042 -0.017 0.020(0.061) (0.063) (0.032)Boric Support 0.885***(0.029)Indigeneity -0.007 0.0003 0.007(0.064) (0.065) (0.033)Abortion Opinion -0.259*** -0.245*** -0.016(0.044) (0.045) (0.024)Constant -1.300*** -1.076*** 0.075(0.280) (0.287) (0.151)N 338 338 338Adj. R2 0.638 0.619 0.902Note: *p < 0.1 **p < 0.05 ***p < 0.01. All variables are standardized. Region fixed effects areincluded in all specifications.Aguilar 40Table 4: Individual-Level Mediation Analysis ResultsSupport forBoricSupport for DraftConstitution(1) (2) (3)Crime Concern -0.900*** -0.883*** -0.380(0.243) (0.245) (0.339)Poverty Measure -0.007 -0.019 -0.059(0.085) (0.083) (0.118)Boric Support 4.375***(0.392)Indigeneity 0.015 0.262 0.369(0.353) (0.343) (0.457)Abortion Opinion -1.662*** -1.842*** -1.366***(0.224) (0.222) (0.301)Age 0.014** -0.001 -0.025***(0.007) (0.006) (0.010)Sex 0.050 -0.016 -0.055(0.217) (0.212) (0.294)Constant -0.251 1.903* 1.928(1.067) (0.990) (1.264)N 516 516 516Log Likelihood -281.401 -294.689 -174.659Note: *p < 0.1 **p < 0.05 ***p < 0.01. Region fixed effects are included in all specifications.The magnitudes of coefficients are not readily interpretable due to the Poverty Measure’s use of aseven-point scale rather than a binary variable.Together, these results suggest that crime is indirectly correlated with the result of the2022 Constitutional Plebiscite. The coefficients of crime are significant in Column 2 of Tables 3Aguilar 41and 4, and they disappear once adding controlling for Boric support in Column 3. Thestatistically significant coefficients of crime in Column 1 of both tables highlight the status ofBoric support as a mediator for crime concerns. In fact, Baron & Kenny (1986) refer to theindependent variable losing support when controlling for the mediator variable, which occurs inColumn 3 of both tables, as “perfect mediation.” The pattern of significance outlined by Tables 3and 4, despite the different directions of the coefficients of the mediator and independentvariable, provides evidence in favor of the notion that support for President Boric acted as anintermediary for the relationship between crime concerns and reduced support for the DraftConstitution. This result remains robust to the use of inverted education level as an alternativemeasure for individual-level poverty (results presented in Table A4).B. The Role of ProtestsI also find a separate and significantly negative relationship between a municipality’s protestactivity and support for the Draft Constitution. Because protests could have contributed tovoters’ perception regarding the document and the threat of future violence, I add a variablecapturing the magnitude of the social unrest in each municipality. This municipal-level data on“contentious events” is collected by the Center for Social Conflict and Cohesion Studies(COES). To measure the magnitude of the social unrest in each municipality, I use the combinedsum of contentious events in 2019 and 2020, as this roughly aligns with the height of the SocialUnrest. I scale this incidence of contentious events in each municipality to incidents per onehundred thousand residents.7 Additionally, I include an interaction variable between this scaledvariable and poverty to examine whether protests shaped the relationship between poverty andapproval of the Draft Constitution. The results are presented in Table 5.7 Four municipalities (Ollagüe, Tortel, Guaitecas, and Contulmo) were removed as outliers. A municipality’spopulation is obtained through the 2017 Census.Aguilar 42Table 5: Municipal-Level Regressions including Contentious EventsSupport for the Draft Constitution(1) (2) (3) (4)2019-2022 Contentious Events 0.002 -0.072 0.001 -0.055**(0.046) (0.049) (0.042) (0.022)Poverty Proportion -0.465*** -0.380*** -0.004 0.004(0.049) (0.058) (0.063) (0.032)Content. Event-Poverty Interaction -0.009 -0.024 -0.070*** 0.002(0.032) (0.031) (0.025) (0.013)Three-Year Change in Crime -0.241*** -0.198*** -0.115*** -0.009(0.046) (0.045) (0.038) (0.020)Crime-Poverty Interaction -0.022 -0.020 -0.046 -0.026*(0.036) (0.035) (0.029) (0.015)Indigeneity -0.037 0.073 0.038(0.058) (0.070) (0.036)Abortion Opinion -0.298*** -0.239*** -0.018(0.049) (0.047) (0.025)Boric Support 0.891***(0.030)Constant 0.018 0.010 -1.004*** 0.187(0.046) (0.044) (0.297) (0.157)N 338 333 333 333Adjusted R2 0.305 0.376 0.634 0.904Note: *p < 0.1 **p < 0.05 ***p < 0.01. Variables are standardized.Aguilar 43The results suggest that protest incidence is associated with support for the DraftConstitution in a manner separate from support. Coefficients of the scaled contentious eventvariable were insignificant in Columns 1, 2, and 3. However, once holding support for PresidentBoric constant in Column 4, I obtain a significant and negative coefficient of the contentiousevents variable. Municipalities with similar levels of Boric support may have been swayed byprotests to vote against the Draft Constitution. The significance and direction of this coefficientprovides evidence in support of Sazo (2023), who suggests that citizens observing the destructionof public spaces became increasingly concerned about the social upheaval. For reference, themarginal effect (-0.055) of this standardized variable is less than a quarter of the marginal effectof Abortion Opinion in Column 3. Its significance nonetheless suggests that protests may haveserved as a noteworthy consideration outside of traditional political cues. Of note, the coefficientof the interaction variable is significant and negative in Column 3, suggesting that contentiousevents may have shaped the relationship between poverty and support for the Draft Constitutionbefore controlling for Boric support. This result may provide limited support for the analysis ofSomma et al. (2021), who argue that low-income individuals bore the burden of violent protests.Overall, while not explicitly captured by measures of crime, protests may have decreased generalvoter support for the Draft Constitution in their own manner.C. Role of the Separate Indigenous Justice SystemI further explore whether the highly salient establishment of a separate Indigenous justice systemresulted in tangible voting differences, yet I do not obtain conclusive results. Using theindividual-level data, I run the original four specifications with an added interaction variablebetween a respondent’s crime concern and their identification as Indigenous. I choose to focusexclusively on individual-level data because the municipal-level data employs actual crime ratesAguilar 44as a proxy rather than using an explicit measure of crime concern. I expect a negative coefficientof the interaction variable, as Indigenous individuals may trust the efficacy of the separate justicesystem more than their non-Indigenous counterparts. Rather than simply regressing crimeconcern against Indigeneity, employing an interaction variable gives insight into the effect ofdraft’s content. Indigenous voters may be similarly influenced by crime concerns to reject theDraft Constitution, but a disconnect in the marginal effect of crime concern between Indigenousand non-Indigenous voters suggests different mechanisms between crime concerns and decreasedsupport for the Draft Constitution. The results are presented in Table 6.Table 6: Individual-Level Regressions with Indigenous-Crime Interaction VariableSupport for the Draft Constitution(1) (2) (3) (4)Indigeneity 0.039 -0.558 -0.504 -0.809(0.641) (0.704) (0.739) (0.946)Crime Concern -2.913*** -2.532*** -2.469** -1.016(0.876) (0.948) (0.963) (1.487)Crime-Indigenous Interaction 0.202 0.870 0.934 1.535(0.723) (0.795) (0.831) (1.079)Poverty Measure -0.373** -0.248 -0.242 -0.128(0.147) (0.159) (0.163) (0.250)Crime-Poverty Interaction 0.390** 0.307* 0.294 0.089(0.167) (0.180) (0.182) (0.276)Abortion Opinion -1.836*** -1.848*** -1.407***(0.214) (0.225) (0.303)Boric Support 4.384***(0.395)Aguilar 45Age -0.007 -0.001 -0.001 -0.025***(0.006) (0.006) (0.006) (0.010)Sex 0.137 -0.058 -0.019 -0.059(0.187) (0.205) (0.212) (0.295)Constant 2.474*** 3.046*** 3.106** 2.373(0.846) (0.918) (1.234) (1.746)N 516 516 516 516Log Likelihood -339.268 -298.382 -292.441 -173.477Note: *p < 0.1 **p < 0.05 ***p < 0.01. The magnitudes of the coefficients are not readilyinterpretable due to the poverty measure’s use of a seven-point scale rather than a binaryvariable.The lack of significance in all specifications for both the Indigenous and theCrime-Indigenous Interaction variables leaves the validity of this mechanism unclear. WhileIndigenous voters may have indeed been less concerned about the establishment of a separateIndigenous justice system contributing to rising crime, there may exist other factors that opposethis effect. For example, the ongoing conflict between Mapuche activists and police in the LaAraucania region may have led Indigenous respondents to feel more uneasy regarding futureviolence. The Draft Constitution may have increased uncertainty regarding this violence,especially given Indigenous groups’ distrust of the government (Piscopo & Siavelis 2023).D. Poverty and Social ConservatismFinally, although not explicitly related to crime concerns, I explore whether the unexpectednegative relationship between poverty and support in the individual-level data is partially shapedby low-income voters’ opinions on the reproductive rights included in the Draft Constitution.Gilens (2012) finds that, in the United States, affluent voters are more liberal on social issuesthan low-income voters. Thus, he argues that differing social and economic preferences acrossAguilar 46income groups are “canceled out” when combined into broader ideologies (Gilens 2012). Itherefore include an interaction variable to examine whether views on abortion are associatedwith a convergence in behavior between low- and high-income voters. Similar to the previoustest of Indigeneity and Crime Concern, I add Abortion Opinion and its interaction with poverty tothe baseline regression. This test, however, uses both datasets. I subsequently add the otherrelevant controls in the following specifications to isolate its effect. The municipal-level resultsare presented below in Table 7, and the individual-level results are presented in Table 8.Table 7: Municipal-Level Regressions with Poverty-Abortion Opinion Interaction VariableSupport for the Draft Constitution(1) (2) (3) (4)Abortion Opinion -0.266*** -0.282*** -0.239*** -0.006(0.046) (0.047) (0.046) (0.025)Poverty Measure -0.404*** -0.368*** -0.004 0.012(0.050) (0.057) (0.063) (0.033)Poverty-Abortion Interaction -0.019 -0.017 -0.070*** 0.004(0.030) (0.030) (0.025) (0.013)Crime Concern -0.201*** -0.202*** -0.115*** -0.009(0.045) (0.045) (0.038) (0.020)Crime-Poverty Interaction -0.017 -0.021 -0.046 -0.028*(0.035) (0.035) (0.029) (0.015)Indigeneity -0.070 0.073 0.031(0.054) (0.070) (0.036)Boric Support 0.884***(0.030)Constant 0.010 0.008 -1.003*** 0.088Aguilar 47(0.044) (0.044) (0.289) (0.154)N 333 333 333 333Adjusted R2 0.373 0.374 0.635 0.902Note: *p < 0.1 **p < 0.05 ***p < 0.01. All variables are standardized.Table 8: Individual-Level Regressions with Poverty-Abortion Opinion Interaction VariableSupport for the Draft Constitution(1) (2) (3) (4)Abortion Opinion -3.921*** -3.972*** -4.001*** -3.399***(0.881) (0.887) (0.900) (1.238)Poverty Measure -0.563*** -0.580*** -0.569*** -0.497(0.205) (0.208) (0.211) (0.320)Poverty-Abortion Interaction 0.419** 0.430** 0.434** 0.401*(0.167) (0.169) (0.171) (0.235)Crime Concern -2.763*** -2.770*** -2.696*** -1.578(1.009) (1.011) (1.029) (1.569)Crime-Poverty Interaction 0.369** 0.369** 0.355* 0.222(0.187) (0.187) (0.190) (0.286)Indigeneity 0.209 0.316 0.466(0.327) (0.337) (0.456)Boric Support 4.344***(0.393)Age -0.002 -0.002 -0.002 -0.025***(0.006) (0.006) (0.006) (0.010)Sex -0.078 -0.080 -0.052 -0.087(0.206) (0.206) (0.214) (0.295)Aguilar 48Constant 4.651*** 4.701*** 4.767*** 4.237***(1.179) (1.186) (1.448) (2.052)N 516 516 516 516Log Likelihood -295.783 -295.579 -289.705 -173.005Note: *p < 0.1 **p < 0.05 ***p < 0.01. The magnitudes of coefficients are not readilyinterpretable due to the Poverty Measure’s use of a seven-point scale rather than a binaryvariable.The results of the two approaches are mixed. As discussed in the two mainmethodologies, favoring abortion restrictions is significantly and negatively associated withsupport for the Draft Constitution in both datasets. However, the two datasets provide contrastingresults regarding the coefficient of the interaction variable between poverty and abortion. Themunicipal-level coefficient of the interaction is significant and negative in Column 3, consistentwith the notion that greater religiosity decreased the marginal effect of poverty on support for theDraft Constitution. The individual-level coefficient, on the other hand, is significant and positivein Columns 1, 2, and 4. This direction suggests a counterintuitive relationship between abortionopinion, poverty, and the Draft Constitution. The negative coefficient of poverty in the first twospecifications, when combined with the positive coefficient of the interaction variable, suggeststhat favoring abortion restrictions is associated with a convergence in voting behavior betweenlow- and high-income voters. This result is consistent with Gilens (2012)’s notion that thegrouping of social and economic preferences tends to “cancel out” differences between incomegroups. The negative direction of the poverty measure, however, suggests that lower-incomevoters were distrustful of the economic redistribution in the first place. Therefore, although thetwo preferences seem to mitigate each other to some degree, they do so from the invertedpredicted directions. The contrasting results from the municipal- and individual-level regressionsleave this question inconclusive.Aguilar 49VII. DiscussionThe results presented provide valuable insight into the dynamics behind the 2022 ChileanConstitutional Plebiscite. Although the government initiated the constitutional process to addressthe economic concerns raised by the 2019 Social Unrest, external factors appear to have shapedthe public’s support of the resultant draft in several unexpected ways. Support for PresidentBoric appears to have played a central role in the plebiscite, and concerns related to rising crimewere mediated through that support to determine the fate of the Draft Constitution. This studydoes not find strong evidence regarding the relationship between poverty and support for theDraft Constitution. These results suggest that the Draft Constitution struggled to gain the supportof those whom it aimed to benefit.Hypothesis 1, which posits that rising crime is associated with decreased support for theDraft Constitution, is largely supported by the data. In both the municipal- and individual-leveldatasets, crime concerns are significantly and negatively associated with support for the DraftConstitution. However, this relationship loses significance once controlling for Boric support. Ithus treat Boric support as an intermediate endogenous variable to examine how crime andpoverty indirectly affect support for the Draft Constitution. Mediation analysis reveals that crimeconcerns are negatively and significantly associated with support for President Boric, which is inturn positively associated with support for the Draft Constitution. Therefore, crime concerns mayhave been associated with decreased support for the Draft Constitution, although primarilythrough voters’ opinion of President Boric. This evidence is consistent with Hagerty (2006)’s“Responsibility Hypothesis” and the findings of Berg Bilbao & Navia (2023). The length of thedraft likely forced voters to rely on cues rather than the concrete articles included in the DraftConstitution (Berg Bilbao & Navia 2023). Lupia (1994) establishes that this effect occurs in priorAguilar 50referendums, and the 2022 Plebiscite may have served as an extension of the trend. I additionallytest whether the establishment of a separate Indigenous justice system, an example of contentwithin the draft, served as a mechanism through which the draft itself was rejected based oncrime concerns. A voter’s Indigenous identity, which I predicted to mitigate the negativerelationship between crime concerns and decreased approval, did not have a noticeable effect.The inconclusive results regarding this mechanism, combined with the strong significanceassociated with Boric support, suggests that opinions of public figures played a pivotal role in therelationship between crime concerns and the Draft Constitution.Hypothesis 2, on the other hand, does not find support in the data. This hypothesis,drawing on the Meltzer-Richard Model, predicted that a higher degree of poverty would beassociated with increased support for the redistributive Draft Constitution when holding constantother second-dimension considerations. Both the municipal- and individual-level datasets find aweakly significant and negative relationship between poverty and support before adding regionfixed effects and controlling for Boric support, providing further evidence against the validity ofHypothesis 2. While these results contradict the notion that crime concerns deter low-incomevoters from otherwise voting for economic redistribution, they serve as a reminder of thecomplexity of the low-income vote. Lingering political elitism may help to understand the failureof this effort towards economic redistribution. As discussed by Luna (2018), many residents inlow-income neighborhoods feel that government institutions have both robbed and abandonedthem. Velasco & Funk (2020) even argue that declining trust in institutions, rather than broaderfrustration with economic inequality, sparked the 2019 Social Unrest. Specifically, voters losttrust in institutions to address the harms of inequality (Velasco & Funk 2020). A study conductedby Piscopo & Siavelis (2023) finds that Indigenous voters rejected the Draft Constitution, despiteAguilar 51its expanded rights for Indigenous communities, precisely because the administration of thoserights fell under the purview of a distrusted government. This distrust may have clashed with thepromise of economic redistribution, ultimately alienating low-income voters. Luna (2018)suggests that the Chilean left historically struggles to understand the demands of working-classvoters. This study does not necessarily confirm this alternate explanation, yet it nonethelessdemonstrates the importance of considering second-dimension considerations to theMeltzer-Richard Model. This inconclusive relationship between income and support for the DraftConstitution reminds public figures, even if they are independent and new to politics, to embracea complex and nuanced understanding of the preferences of low-income voters.This study largely does not find any evidence supporting Hypothesis 3, which aimed toexplore whether crime concerns shaped the relationship between a voter’s low-income status andtheir support of the Draft Constitution. The municipal-level regression outlined in Table 1 findslimited evidence for the negative interaction variable once controlling for Boric support, yet thisidea is not confirmed by the individual-level regressions. Before adding region fixed effects andcontrolling for Boric support, both the municipal- and individual- level regressions find analready-existing negative relationship between poverty and support for the Draft Constitution.The original prediction regarding the negative coefficient of the interaction variable betweenpoverty and crime concern was based on the assumption that the relationship between povertyand support would be positive. The failure of Hypothesis 2, however, does not support thisassumption. Low-income voters may therefore have distrusted the Constitution regardless ofcrime concerns.Aguilar 52VIII. ConclusionThe widespread public outcry regarding the lingering harmful effects of the 1980 Constitutioncentered the notion of economic inequality. However, as the nation moved closer to voting on theDraft Constitution, voters expressed their discontent with the harmful effects of crime. Thisstudy sought to explore whether low-income voters’ exposure to crime ultimately led them toreject the Constitution despite its redistributive elements.Ultimately, this study finds that crime is associated with decreased support for the DraftConstitution specifically through decreased support for President Boric. The importance of Boricsupport in the 2022 Plebiscite is consistent with studies conducted by Berg Bilbao & Navia(2023) and Gonzalez Ocantos & Melendez (2023). However, this study does not find theexpected positive relationship between poverty and support for the Draft Constitution whencontrolling for salient second-dimension considerations. The counter-intuitive result, given thedraft’s attempts to address the economic and political demands of low-income voters, mayindicate that low-income voters continued to perceive the Draft Constitution as the product of adistrusted political elite. While the constitutional process appears to have been put on hold afterthe rejection of the second, more conservative Draft Constitution (Al Jazeera 2023), this insightmay be useful in future attempts to address public discontent.Limitations of this study may have impacted the validity of the results. For example, dueto a lack of data regarding economic inequality within each commune, I relied on the percentageof residents living in income-based poverty as a proxy for the proportion of residents who endurethe harmful effects of income inequality. I encourage future studies to make use of futuremunicipality-level inequality data if it becomes available. Similarly, using the municipality as theunit of analysis may not fully capture the mindsets of individual voters. The individual-levelAguilar 53models seek to mitigate this concern, yet future research into the behavior of individual votersusing government-collected data may yield more precise information. One could also argue thatchanges in crime may have modified broader political beliefs rather than support for PresidentBoric himself, and I suggest that future projects should explore this distinction.Overall, this study highlights the complex factors that inform the voting decisions oflow-income citizens. Political elitism may have sparked the protests, but it may have similarlyprevented the approval of a new constitution. Political actors may benefit from the reminder toactively root themselves among those whose interests they aim to represent. Actively listeningand expanding perspectives can facilitate well-informed initiatives that create meaningful,sustainable change.Aguilar 54IX. AcknowledgementsI am very grateful to my faculty advisor, Professor Vicky Murillo, who provided me withinvaluable knowledge and guidance throughout this process. This thesis also greatly benefitedfrom the continued support of Professor John Huber and Martin Devaux. I would also like tothank Professors Patricio Navia, Carla Alberti, and Nicolás Somma for their insights. Finally, Iwish to acknowledge my amazing family and friends for their words of encouragementthroughout this new experience.X. BibliographyAguirre, Jerjes, and Amador Herrera, Hugo. 2013. “Institutional Weakness and Organized Crimein Mexico: the Case of Michoacán.” Trends in Organized Crime 16: 221-238.Alemán, Eduardo, and Navia, Patricio. 2023. Chile’s Failed Constitution: Democracy Wins.Journal of Democracy 34(2): 90–104.Alemán, Eduardo, and Saiegh, Sebastián. 2007. “Legislative Preferences, Political Parties, andCoalition Unity in Chile.” Comparative Politics 39(3): 253–272.http://www.jstor.org/stable/20434040Ansaldi, Octavio, and Pardo-Vergara, Maria. 2020. “What Constitution? On Chile’sConstitutional Awakening.” Law and Critique 31: 7-39.“Apruebo o Rechazo: 6 Grandes Cambios que Propone la Nueva Constitución que Vota Chile.”BBC News Mundo. https://www.bbc.com/mundo/noticias-america-latina-62010439(December 17, 2023).Baer, Madeline. 2014. “Private Water, Public Good: Water Privatization and State Capacity inChile.” Studies in Comparative International Development, 49: 141-167.Baron, Reuben, & Kenny, David. 1986. “The Moderator–Mediator Variable Distinction in SocialPsychological Research: Conceptual, Strategic, and Statistical Considerations.” Journalof Personality and Social Psychology, 51(6): 1173.Barton, Jonathan. 2014. Chile. The South America Handbook: 99-108.Belmar, Fabián, Morales, Mauricio, & Villarroel, Benjamín. 2023. “Writing a Constitutionwithout Parties? The Programmatic Weakness of Party-Voter Linkages in the ChileanPolitical Change.” Politics: 1-21.Aguilar 55Berg Bilbao, Santiago, and Navia, Patricio. 2023. “I Fear crime, I Vote against the Government.The Perception of Insecurity and Vote Choice in Constitutional Plebiscites.” WorkingPaper.Berstein, Solange, Larraín, Guillermo, Pino, Francisco, and Morón, Eduardo. 2006. “ChileanPension Reform: Coverage Facts and Policy Alternatives [with Comments].” Economía6(2): 227–279.Biroli, Flávia. 2020. “The Backlash against Gender Equality in Latin America: Temporality,Religious Patterns, and the Erosion of Democracy.” Lasa Forum 51(2): 22-26.Bonner, Michelle. 2013. “The Politics of Police Image in Chile.” Journal of Latin AmericanStudies 45(4): 669-694.Borzutzky, Silvia. 2005. “From Chicago to Santiago: Neoliberalism and Social SecurityPrivatization in Chile.” Governance 18(4): 655-674.Brender, Valerie. 2010. “Economic Transformations in Chile: The Formation of the ChicagoBoys.” The American Economist 55(1): 111-122.Briones, Sebastián, and Dockendorff, Andrés. 2015. “Continuidad y Cambio en la PolíticaExterior Chilena en el Gobierno de Sebastián Piñera (2010-2014).” EstudiosInternacionales (Santiago) 47(180): 115-138.Brunner, José Joaquín. 2021. “La Rebelión de una Generación Desengañada.” La Revuelta deOctubre en Chile: 75-104.Castiglioni, Rossana. 2021. “Políticas Sociales, Precariedad y Malestar Social en Chile.” LaRevuelta de Octubre en Chile: 105-127.Castillo, Juan, Palacios, Diego, Joignant, Alfredo, and Tham, Maximiliano. 2015. “Inequality,Distributive Justice and Political Participation: An Analysis of the Case of Chile.”Bulletin of Latin American Research 34(4): 486-502.Castillo-Laborde, Carla, Aguilera-Sanhueza, Ximena, Hirmas-Adauy, Macarena, Matute, IsabelDelgado-Becerra, Iris, Ferrari, Manuel, Olea-Normandin, Andrea, andGonzález-Wiedmaier, Claudia. 2017. “Health Insurance Scheme Performance and Effectson Health and Health Inequalities in Chile.” MEDICC Review 19: 57-64.“Censo de Población y Vivienda.” Instituto Nacional de Estadísticas Chile.http://www.ine.gob.cl/estadisticas/sociales/censos-de-poblacion-y-vivienda/censo-de-poblacion-y-vivienda (February 26, 2024).Aguilar 56Center for Social Conflict and Cohesion Studies (COES). “Observatory of Conflicts -Cumulative Dataset.”https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/GKQXBR.“Chile’s Constitutional Assembly Votes to Replace Senate with New Chamber.” 2022. Reuters.https://www.reuters.com/world/americas/chiles-constitutional-assembly-votes-replace-senate-with-new-chamber-2022-04-14/ (February 17, 2024).“Chile’s Constitutional Convention: A Triumph of Inclusion.” UNDP.https://www.undp.org/latin-america/blog/chiles-constitutional-convention-triumph-inclusion (February 17, 2024).Corvalan, Alejandro, and Cox, Paulo. 2013. “Class-Biased Electoral Participation: The YouthVote in Chile.” Latin American Politics and Society 55(3): 47–68.http://www.jstor.org/stable/43284847Cullen, Francis, Wright, John Paul, and Chamlin, Mitchell 1999. “Social Support and SocialReform: A Progressive Crime Control Agenda.” Crime & Delinquency 45(2): 188-207.De La O, Ana L., and Rodden, Jonathan. 2008. “Does Religion Distract the Poor? Income andIssue Voting around the World.” Comparative Political Studies 41(4-5): 437-476.Di Tella, Rafael, Freira, Lucía, Gálvez, Ramiro, Schargrodsky, Ernesto, Shalom, Diego, andSigman, Mariano. 2019. “Crime and Violence: Desensitization in Victims to WatchingCriminal Events.” Journal of Economic Behavior & Organization 159: 613-625.Disi Pavlic, Rodolfo. 2024. “Indigenous Voices and Votes: Assessing the Dynamics ofIndigenous Politics in Chile’s Constitutional Referendum of 2022.” PS: Political Science& Politics: 1-7.“Does Inequality Matter?: How People Perceive Economic Disparities and Social Mobility. HowDoes Chile Compare?” 2021. Organization for Economic Cooperation and Development.https://www.oecd.org/wise/Does-inequality-matter-country-note-Chile.pdf.Dong, Baomin, Egger, Peter, and Guo, Yibei. 2020. “Is Poverty the Mother of Crime? Evidencefrom Homicide Rates in China.” PloS one 15(5): e0233034.“Elección Presidencial 2021.” Servicio Electoral de Chile.https://www.servel.cl/centro-de-datos/resultados-electorales-historicos-gw3/.“Encuesta CEP N° 88, Noviembre-Diciembre 2022.” 2023. Centro de Estudios Públicos.https://www.cepchile.cl/encuesta/encuesta-cep-n-88/.“Encuesta de Ipsos y Espacio Público por Proceso Constituyente: 74% piensa que el paísnecesita una Nueva Constitución – Espacio Público.”Aguilar 57https://espaciopublico.cl/encuesta-de-ipsos-y-espacio-publico-por-proceso-constituyente-74-piensa-que-el-pais-necesita-una-nueva-constitucion/ (February 28, 2024).Escudero, María Cristina, and Olivares, Alejandro. 2021. “From the Dance of Those Left Out toa New Constitution: Channeling the Chilean Social Unrest.” The Social Outburst andPolitical Representation in Chile: 173-192.“Estadísticas Delictuales.” Portal CEAD. https://cead.spd.gov.cl/estadisticas-delictuales/(February 26, 2024).“Estudio Nacional de Opinión Pública N°86, Abril-Mayo 2022.” 2022. Centro de EstudiosPúblicos.https://www.cepchile.cl/encuesta/estudio-nacional-de-opinion-publica-n86-abril-mayo-2022/.“Eyes on Chile.” 2020. Amnesty International.https://www.amnesty.org/en/latest/research/2020/10/eyes-on-chile-police-violence-at-protests/ (February 19, 2024).Fay III, Robert., and Herriot, Roger. 1979. “Estimates of Income for Small Places: anApplication of James-Stein Procedures to Census Data. Journal of the AmericanStatistical Association 74(366a): 269-277.Feldmann, Andreas, and Luna, Juan Pablo. 2023. “Criminal Politics and Botched Developmentin Contemporary Latin America.” Elements in the Politics of Development.Fuentes, Claudio. 2015. “Shifting the Status Quo: Constitutional Reforms in Chile.” LatinAmerican Politics and Society 57(1): 99-122.Fuentes, Claudio. 2021. “La Dinámica Sociopolítica del Proceso Constituyente en Chile.” LaRevuelta de Octubre en Chile: 129-153.Fuller, Will. 2024. “Chileans Still Want a New Constitution, Now They Must Decide Who WillWrite It.” International Relations Review.https://www.irreview.org/articles/chileans-still-want-a-new-constitution-now-they-must-decide-who-will-write-it (February 24, 2024).Gilens, Martin. 2012. Affluence and influence: Economic inequality and political power inAmerica. Princeton University Press.“Gini Index - Chile.” The World Bank.https://data.worldbank.org/indicator/SI.POV.GINI?locations=CL.Gillette, Clayton P. 1988. “Plebiscites, Participation, and Collective Action in Local GovernmentLaw.” Michigan Law Review 86(5): 930–988.Aguilar 58“Global Peace Index Map: The Most & Least Peaceful Countries.” Vision of Humanity.https://www.visionofhumanity.org/maps/ (March 31, 2024).Gonzalez Ocantos, Ezequiel, and Carlos Melendez. 2023. “Rethinking the Role of Issue-Votingin Referenda: Conjoint and Vote Choice Analyses of Preferences for ConstitutionalChange in Chile.” Comparative Politics.Hadzi-Vaskov, Metodij, and Ricci, Luca Antonio. 2021. “Understanding Chile’s Social Unrest inan International Perspective (No. 16494).” International Monetary Fund.Hagerty, Michael. 2006. “Quality of Life from the Voting Booth: the Effect of Crime Rates andIncome on Recent US Presidential Elections.” Social Indicators Research 77: 197-210.Han, Kyung Joon. 2016. “Income Inequality and Voting for Radical Right-Wing Parties.”Electoral Studies 42: 54-64. Hanslmaier, Michael. (2013). Crime, fear and subjective well-being: How victimization andstreet crime affect fear and life satisfaction. European Journal of Criminology, 10(5):515-533.Heiss, Claudia. 2021. “Latin America Erupts: Re-Founding Chile.” Journal of Democracy 32(3):33-47.“How a Pokémon Character Is Rewriting Chile’s Future.” The Economist.https://www.economist.com/1843/2021/12/02/how-a-pokemon-character-is-rewriting-chiles-future (February 20, 2024).Iversen, Torben, and Goplerud, Max. 2018. “Redistribution without a Median Voter: Models ofMultidimensional Politics.” Annual Review of Political Science 21: 295-317.Jasiewicz, Krzysztof. 2009. “The Past Is Never Dead: Identity, Class, and Voting Behavior inContemporary Poland.” East European Politics and Societies 23(04): 491-508.Kapstein, Ethan, and Tantravahi, Adityamohan. 2021. “The Economic Consequences of CriminalViolence.” Political Violence at a Glance.https://politicalviolenceataglance.org/2021/10/11/the-economic-consequences-of-criminal-violence/ (February 22, 2024).Krahn, Harvey, Hartnagel, Timothy, and Gartrell, John. 1986. “Income Inequality and HomicideRates: Cross‐National Data and Criminological Theories.” Criminology 24(2): 269-294.Laakso, Markku, and Taagepera, Rein. 1979. “‘Effective’ Number of Parties: a Measure withApplication to West Europe.” Comparative Political Studies 12(1): 3-27.Aguilar 59Lacoe, Johanna, Bostic, Raphael, and Acolin, Arthur. 2018. “Crime and Private Investment inUrban Neighborhoods.” Journal of Urban Economics 108: 154-169.Lofstrom, Magnus, and Raphael, Steven. 2016. “Crime, the Criminal Justice System, andSocioeconomic Inequality.” Journal of Economic Perspectives 30(2): 103-126.Luna, Juan Pablo. 2018. En Vez del Optimismo: Crisis de Representación Política en el ChileActual. Editorial Catalonia.Lupia, Arthur. 1994. “Shortcuts versus Encyclopedias: Information and Voting Behavior inCalifornia Insurance Reform Elections.” American Political Science Review 88(1): 63-76.Matear, Ann. 2007. “Equity in Education in Chile: The Tensions between Policy and Practice.”International Journal of Educational Development 27(1): 101-113.Meltzer, Allan, & Richard, Scott. 1981. “A Rational Theory of the Size of Government.” Journalof Political Economy 89(5): 914–927. http://www.jstor.org/stable/1830813Ministerio de Desarrollo Social y Familia, and CEPAL. “Estimaciones Comunales de PobrezaPor Ingresos En Chile Mediante Métodos de Estimación En Áreas Pequeñas.”Morales Quiroga, Mauricio. 2021. “Chile’s Perfect Storm: Social Upheaval, COVID-19 and theConstitutional Referendum.” Contemporary Social Science 16(5): 556-572.Motamedi, Maziar. “Why Chileans Rejected Conservative Constitution, and What’s next?” AlJazeera.https://www.aljazeera.com/news/2023/12/18/why-chileans-rejected-conservative-constitution-and-whats-next (February 25, 2024).Newman, Lucia. “A Journey through Chile’s Conflict with Mapuche Rebel Groups.” Al Jazeera.https://www.aljazeera.com/features/2021/4/12/a-journey-through-chiles-conflict-with-mapuche-resistance-groups (February 18, 2024).Palanza, Valeria, and Valarezo, Patricia. 2023. “Chile’s Failed Constitutional Intent: Polarization,Fragmentation, Haste and Delegitimization.” Global Constitutionalism: 1-10.Parra Coray, Gonzalo. 2021. “Political Parties and Social Movements in Post-transition Chile:Between Mistrust and Reconfiguration.” The Social Outburst and PoliticalRepresentation in Chile: 51-74.Peña, Carlos. 2021. “El Malestar en la Modernización: El Caso Chileno.” La Revuelta deOctubre en Chile: 19-46.Aguilar 60Peña, Carlos, and Silva, Patricio. 2021. “Más Allá de la Revuelta: El Chile que Viene.” LaRevuelta de Octubre en Chile: 155-167.“Pinochet’s Ghost Haunts Divisive Chile Election.” Reuters.https://www.reuters.com/world/americas/pinochets-ghost-haunts-divisive-chilean-election-2021-11-21/ (December 17, 2023).Piscopo, Jennifer, and Siavelis, Peter. 2023. “Chile's Constitutional Chaos.” Journal ofDemocracy 34(1): 141-155.“Plebiscito Constitución Política 2020.” Servicio Electoral de Chile (ServEl).https://app.powerbi.com/view?r=eyJrIjoiN2Y0ODM3MmUtZmY3YS00N2ZjLWJjNjMtM2Y4MjU3Y2UyNjEzIiwidCI6ImVhZjg3OWJkLWQzZWMtNDY1MC1iMTI5LTEzZGZkZjQ4NTlmZSJ9.“Plebiscito Constitucional 2022.” Servicio Electoral de Chile (ServEl).https://app.powerbi.com/view?r=eyJrIjoiNDI2ZWYxNmItYWIwNC00YzYzLWE1OTUtZTZhNjE5MWYzNDdlIiwidCI6ImVhZjg3OWJkLWQzZWMtNDY1MC1iMTI5LTEzZGZkZjQ4NTlmZSJ9“Propuesta: Constitución Política de La República de Chile.” 2022.Putnam, Kate Marie. 2014. “Voting Behavior in Violence-Plagued New Democracies: CrimeVoting in Mexico’s Recent Presidential Elections” (Doctoral dissertation).“¿Qué Dice La Propuesta Constitucional Sobre Seguridad Pública?” La Neta. 2022.https://laneta.cl/que-dice-la-propuesta-constitucional-sobre-seguridad-publica/#:~:text=El%20art%C3%ADculo%2053%20del%20texto,derecho%20a%20todas%20las%20personas.Rahat, Gideon, and Sznajder, Mario. 1998. “Electoral engineering in Chile: The Electoral Systemand Limited Democracy.” Electoral Studies 17(4): 429-442.“Religión Declarada.” https://www.bcn.cl/siit/estadisticasterritoriales/tema?id=101.Rhodes-Purdy, Matthew, and Rosenblatt, Fernando. 2023. “Raising the Red Flag: DemocraticElitism and the Protests in Chile. Perspectives on Politics 21(1): 241-253.Rioja, Romina Green, Loncon, Elisa, and Carrillo Vidal, Alondra. 2022. Transforming Chilefrom the Ground Up: Two Constitutional Convention Representatives Reflect on theProcess of Drafting the World’s Most Progressive Constitution. NACLA Report on theAmericas, 54(4), 382-388.Rivera, Víctor. 2019. “Carabineros Estima 3,7 Millones de Asistentes a Manifestaciones En 24Días.” La Tercera.Aguilar 61https://www.latercera.com/nacional/noticia/carabineros-estima-37-millones-asistentes-manifestaciones-24-dias/896954/ (February 24, 2024).Rueda, David, and Stegmueller, Daniel. 2016. “The Externalities of Inequality: Fear of Crimeand Preferences for Redistribution in Western Europe.” American Journal of PoliticalScience 60(2): 472-489.Sampó, Carolina, and Troncoso, Valeska. 2023. “Cocaine Trafficking from Non-TraditionalPorts: Examining the Cases of Argentina, Chile and Uruguay.” Trends in OrganizedCrime 26(3): 235-257.Sazo, Diego. 2023. “Chile 2022: From Great Expectations to Rising Pessimism.” Revista deCiencia Política 43(2): 193-222Schaffner, Brian, and Senic, Nenad. 2006. “Rights or Benefits? Explaining the Sexual IdentityGap in American Political Behavior.” Political Research Quarterly 59(1): 123–132.http://www.jstor.org/stable/4148080Silva, Patricio. 2021. “La Rebelión de Octubre y sus Raíces Históricas.” La Revuelta de Octubreen Chile: 47-74.Skogan, Wesley. 1986. “Fear of Crime and Neighborhood Change.” Crime and Justice 8:203-229.Somma, Nicolás M. 2017. “Discontent, Collective Protest, and Social Movements in Chile.”Malaise in Representation in Latin American Countries: Chile, Argentina, and Uruguay:47-68.Somma, Nicolás M., Bargsted, Matías, Disi Pavlic, Rodolfo, and Medel, Rodrigo. 2021. “NoWater in the Oasis: The Chilean Spring of 2019–2020.” Social Movement Studies 20(4):495-502.Tricot, Victor. 2021. “Please Mind the Gap: Autonomization and Street Politics.” The SocialOutburst and Political Representation in Chile: 75-89.Velasco, Andrés, and Funk, Robert. 2020. “Institutional Vulnerability, Breakdown of Trust: AModel of Social Unrest in Chile.” Working Paper.Verdugo, Sergio, and Prieto, Marcela . 2021. “The Dual Aversion of Chile’s Constitution-MakingProcess.” International Journal of Constitutional Law 19(1): 149-168.Vergara, Camila. 2022. “Chile’s Rejection.” Politics: 9.Aguilar 62“Voters Should Reject Chile’s New Draft Constitution.” The Economist.https://www.economist.com/leaders/2022/07/06/voters-should-reject-chiles-new-draft-constitution (February 18, 2024).Waissbluth, Mario. 2020. “Orígenes y Evolución del Estallido Social en Chile.” Santiago deChile: Centro de Estudios Públicos Universidad de Chile.Wilenmann, Javier. 2020. “Neoliberal Politics and State Modernization in Chilean PenalEvolution.” Punishment & Society 22(3): 259-280.Yáñez, Bernardo Navarrete. 2021. “Party System Crisis: The Exhaustion of Chile’s Spine.” TheSocial Outburst and Political Representation in Chile: 13-36.Yashar, Deborah. 2018. Homicidal Ecologies: Illicit economies and complicit states in LatinAmerica. Cambridge University Press.Zehr, Howard. 1976. Crime and the Development of Modern Society: Patterns of Criminality inNineteenth Century Germany and France. Routledge.“5 cambios que Propone Hacer el Gobierno de Chile al Proyecto de Constitución si se Apruebael Texto.” BBC News Mundo.https://www.bbc.com/mundo/noticias-america-latina-62511162 (December 17, 2023).Aguilar 63XI. AppendixA. Municipal-Level AnalysisTable A1: Municipal-Level Descriptive StatisticsVariable Source Count Mean Min Max St. Devpercent_approve ServEl 346 31.5% 5.2% 56.7% 9.2%indig_proportionCenso2017346 17.3% 2.1% 85.7% 17.3%poverty_proportionMinisteriodeDesarrolloSocial yFamilia345 13.0% 1.4% 29.0% 4.2%Percent_religiousCenso2002341 93.8% 82.4% 99.3% 2.8%crime_threeyearchangeCEAD 344 4.9% -76.9% 200.0% 30.9%boric_support ServEl 346 20.5% 1.9% 44.9% 7.1%Percent_religious includes Catholic, Jehovah’s Witness, Mormon, Orthodox, Jewish, andMuslim residents. Crimes included in crime_threeyearchange are the following: Sexual abuseand other sexual crime, discovery of bodies and other deaths, homicides, robbery, infractionrelated to gun laws, infraction related to drug laws, injuries (ranging from mild to more serious),receiving stolen property, robbery (with violence or intimidation, from a vehicle, of motorizedvehicle, in inhabited place, in uninhabited place, failed robbery, robbery by surprise), and rape.Excluded crimes are cattle rustling, incivility, domestic violence, and other acts. These crimesare considered less visible to voters. All variables were standardized before being used in theregressions.Aguilar 64B. Individual-Level AnalysisTable A2: Individual-Level Descriptive StatisticsVariable Source Count Mean Min Max St. DevApprove_ConstitutionCEP No.88516 0.44 0 1 0.50Inverted_monthly_incomeCEP No.88516 4.96 1 7 1.35Crime_ConcernCEP No.88516 0.76 0 1 0.43Indigenous CEP No.88516 0.10 0 1 0.31Pro_abortion_restrictionCEP No.88516 0.64 0 1 0.48Inverted_education_levelCEP No.88516 5.07 0 10 2.18Age CEP No.88516 46.95 18 87 16.06Sex CEP No.88516 1.527 1 2 0.50Approve_Constitution is a dummy variable equal to 1 if the respondent rejected thePlebiscite or would have done so if they did not participate. Inverted_monthly_income is a scaleof 1 to 7 that corresponds to different values based on respondents’ reported number ofhousehold members. Income decreases as the scale increases. Inverted_education_level issimilarly ordered as a scale of decreasing levels of education. Crime_concern is equal to 1 if therespondent lists crime, assault, robbery, drugs, and/or violence as one of the top 3 concerns thatthe government should address, and it is equal to 0 otherwise. Indigenous is a dummy variableAguilar 65equal to 1 if the respondent identifies as pertaining to an Indigenous group, and 0 otherwise. Sexis equal to 1 if the respondent identifies as male, and it is equal to 2 if the respondent identifies asfemale.Aguilar 66Table A3: Using Inverted Education Level as Poverty MeasureSupport for the Draft Constitution(1) (2) (3) (4)Crime Concern -1.871*** -1.486** -1.456** -0.138(0.571) (0.620) (0.631) (0.932)Poverty Measure (Inv. Edu.) -0.287*** -0.174* -0.188* -0.032(0.090) (0.097) (0.101) (0.144)Crime-Poverty Interaction 0.191* 0.115 0.116 -0.044(0.102) (0.110) (0.113) (0.160)Indigeneity 0.163 0.282 0.355(0.330) (0.343) (0.456)Abortion Opinion -1.757*** -1.777*** -1.360***(0.213) (0.224) (0.303)Boric Support 4.367***(0.395)Age -0.003 0.001 0.002 -0.023**(0.006) (0.006) (0.007) (0.010)Sex 0.141 -0.039 -0.003 -0.080(0.185) (0.202) (0.209) (0.290)Constant 1.848*** 2.427*** 2.573** 1.724(0.617) (0.683) (1.048) (1.355)N 516 516 516 516Log Likelihood -335.845 -298.633 -292.244 -174.329Note: *p < 0.1 **p < 0.05 ***p < 0.01. The magnitudes of coefficients are not readilyinterpretable due to the alternate Poverty Measure’s use of an eleven-point scale rather than abinary variable.Aguilar 67Table A4: Mediation Analysis using Inverted Education Level as Poverty MeasureSupport forBoricSupport for DraftConstitution(1) (2) (3)Crime Concern -0.873*** -0.865*** -0.375(0.245) (0.247) (0.339)Poverty Measure (Inv. Edu) -0.103** -0.100** -0.066(0.052) (0.051) (0.072)Boric Support 4.355***(0.391)Indigeneity 0.040 0.289 0.347(0.352) (0.344) (0.454)Abortion Opinion -1.599*** -1.791*** -1.354***(0.226) (0.224) (0.302)Age 0.017** 0.002 -0.023**(0.007) (0.007) (0.010)Sex 0.072 -0.005 -0.081(0.215) (0.209) (0.290)Constant 0.026 2.137** 1.902(1.033) (0.950) (1.190)N 516 516 516Log Likelihood -279.448 -292.783 -174.366Note: *p < 0.1 **p < 0.05 ***p < 0.01. Region fixed effects are included in all specifications.The magnitudes of coefficients are not readily interpretable due to the alternate PovertyMeasure’s use of an eleven-point scale rather than a binary variable.

Clean Full Text

Language

Doi

Arxiv

Mag

Acl

Pmid

Pmcid

Pub Date

Pub Year

Journal Name

Journal Volume

Journal Page

Publication Types

Tldr

Tldr Version

Generated Tldr

Search Term Used

Reference Count

Citation Count

Influential Citation Count

Last Update

Status

Aws Job

Last Checked

Modified

Created