Political Psychology, Vol. 0, No. 0, 2021
0162-895X © 2021 International Society of Political Psychology
Published by Wiley Periodicals, Inc., 350 Main Street, Malden, MA 02148, USA, 9600 Garsington Road, Oxford, OX4 2DQ,
and PO Box 378 Carlton South, 3053 Victoria, Australia
The Quest for Uncontested Power: Presidents’ Personalities and
Democratic Erosion in Latin America, 1945– 2012
Ignacio Arana Araya
Carnegie Mellon University
There is a growing scholarly consensus that overreaching heads of government are subverting democracies
across the globe. However, the characteristics of these leaders remain unclear. This article examines a type of
overreaching presidential behavior that has been commonplace in Latin America: between 1945 and 2012, 25
presidents from 14 countries tried to change their respective constitutions to increase their powers. Building
on personality research and semistructured interviews conducted with former presidents, this article proposes
that risk taking and assertive leaders are more likely to try to increase their powers. Using a novel database,
I conduct discrete- time- duration models to test the hypotheses on the presidents that governed from 1945 to
2012. The results demonstrate that the personalities of presidents are a strong force behind their attempts
to consolidate their authority. These findings challenge current approaches in presidential studies and have
implications for the study of all types of political elites.
KEY WORDS: presidents, Latin America, democratic erosion, personalities, constitutional change
The overreaching behavior of national leaders is one of the main factors that explain the current
wave of autocratization— the decay of democratic attributes— that numerous countries are experi-
encing (Lührmann & Lindberg, 2019). There is a growing agreement among scholars that, instead
of perpetrating coups and open electoral frauds, heads of government with authoritarian tendencies
take small steps to consolidate their power (e.g., Levitsky & Ziblatt, 2018; Runciman, 2018). There
are two important challenges for this line of research: to reveal which types of heads of government
erode democratic norms and institutions and the means by which they do so. Scholars have discussed
how leaders have harassed the opposition, undermined judicial and military independence, attacked
the free press, and used state institutions such as taxation and law- enforcement agencies as weap-
ons against their rivals (e.g., Levitsky & Ziblatt, 2018). However, little is known about the personal
characteristics of overreaching leaders. This article starts covering this lacuna, by examining the in-
dividual differences of Latin American presidents who attempted to change the constitutions of their
countries to increase their formal powers.
The expansion of presidential powers has been associated with far- reaching outcomes.
Researchers have claimed that powerful executives are better positioned to lead reforms (e.g.,
Mainwaring & Shugart, 1997) because they face fewer veto players and therefore do not need to
build broad coalitions to govern (Haggard & Kaufman, 1994). Powerful presidencies have also been
linked to an arbitrary use of the office (Linz, 1990) and a worse democratic performance (Helmke,
2017; Shugart & Carey, 1992). Stronger presidents are able to confront or bypass Congress and the
judiciary (Linz, 1990; Shugart & Carey, 1992), and therefore they can isolate themselves from public
pressures. Furthermore, they also tend to make more unilateral and careless policies (Helmke, 2017).
In sum, the leeway strong presidents enjoy comes at the expense of policy stability, electoral repre-
sentation, and checks and balances.
Between 1945 and 2012, 25 Latin American leaders tried 27 times to change the constitution
of their countries to expand their powers. These attempts took place in 14 of the region’s 19 coun-
tries and across all regime types (details are in the appendix in the online supporting information).
The concentration of executive powers helped several leaders to attain historical prominence. Juan
Domingo Perón, Evo Morales, José María Velasco Ibarra, Daniel Ortega, and Hugo Chávez changed
the political paths of Argentina, Bolivia, Ecuador, Nicaragua, and Venezuela, respectively. The be-
havior of these presidents also contributed to a democratic collapse (Alberto Fujimori of Peru) and
was instrumental in the ousting of four presidents (Perón, Gualberto Villarroel of Bolivia, Marcos
Pérez Jiménez of Venezuela, and Jorge Serrano Elías of Guatemala) and the resignation of another
(Juan María Bordaberry of Uruguay). Moreover, an examination of the Polity IV scores (Marshall,
Jaggers, & Gurr, 2014) suggests that presidents who adapt constitutions for self- serving purposes
lead autocratization processes. Of the 21 presidents who increased their powers, eight led govern-
ments that maintained and nine that augmented authoritarianism in their countries.
Despite the frequency and relevance of the presidential attempts to increase their powers, only
two types of arguments have been proposed to explain them. One centers on the institutional distribu-
tion of political power and the other on socioeconomic factors (Armingeon & Careja, 2008; Corrales,
2018; Frye, 1997; Negretto, 2009; Pennings, 2003; Przeworski, 1991; Shugart, 1993).
Previous arguments, nonetheless, have missed a key explanatory factor: the unique character-
istics of presidents. Heads of government have historically been protagonists of the constitutional
changes that have increased their powers. Such patterns of behavior should not be surprising. National
leaders have strong motivations to centralize authority. Doing so helps them to achieve their policy
goals and to overcome gridlocks with Congress. More powers also allow presidents to have tighter
control over the cabinet and, by extension, over the executive branch.
This article proposes that a driving force behind the self- serving constitutional changes are the
personality traits of leaders. It hypothesizes that high risk- taking heads of government are more
likely to bear the uncertainties involved in challenging the political context to pursue self- serving
goals. Presidents who expand their powers gain significant capacity to advance their agenda, but if
they fail in their attempts, they can lose substantial political capital or even be ousted from office (as
the four aforementioned cases illustrate). It is also hypothesized that more assertive leaders are more
likely to try to expand their powers because they are highly motivated to succeed, have outstanding
negotiation skills, and desire to control their context. These hypotheses will be tested using a novel
database of presidents from the Western Hemisphere.
What Explains the Expansion of Presidential Powers?
The literature has advanced two types of arguments to explain increases in presidential pow-
ers. One focuses on the institutional distribution of political power among branches of government
(Corrales, 2018; Frye, 1997; Negretto, 2009, 2013a; Shugart, 1993). The other centers on broad
socioeconomic factors (Armingeon & Careja, 2008; Pennings, 2003; Przeworski, 1991).
Scholars in the first group propose that constitution- makers change the rules expecting that
the new regulations will aid them electorally or will increase their influence over policy outcomes.
Therefore, researchers have argued that the current and expected power asymmetries between legis-
lative parties explain changes in presidential powers. Negretto (2009) argued that parties who control
or expect to control the presidency tend to increase the presidents’ legislative powers when they
The Quest for Uncontested Power
anticipate not having sufficient strength to pass public policies in Congress. He found support for his
theory after examining the main democratic constitutional revisions that took place in Latin America
from 1900 to 2001. Similarly, Corrales (2018) examined 24 constitutional moments in Latin America
since the 1980s and found that power asymmetries between incumbents and opposition parties ex-
plain changes in presidential powers. While powerful heads of government struggle to expand their
powers in a new constitution, opposition constitution- makers try to do the opposite, or at least block
the presidential attempts.
Frye (1997) examined presidential powers in 24 postcommunist countries and found that the
leading parties support the expansion of presidential powers when they have more certainty about
their electoral chances of controlling the presidency. Finally, Shugart (1993) claimed that parties
who dominate the legislature may favor stronger presidents to allow the heads of government to
take direct responsibility for policy choices while legislators focus on providing services to their
Characteristics of other institutions that distribute political power may also explain changes
in presidential powers. Judicial independence may insulate constitutions from overreaching presi-
dents because high courts can interpret charters and check the power of politicians. For example,
Elkins, Ginsburg, and Melton (2009) found that when courts can review the constitutionality of laws,
constitutions last longer because politicians are less likely to violate them. In contrast, when high
courts are not fully independent, they may end up supporting ambitious presidents. That happened
in Argentina, where the Supreme Court helped President Carlos Menem strengthen the executive
(Larkins, 1998). Furthermore, presidents who have strong constitutional powers may be more willing
to take advantage of their bargaining capacity and try to increase them even further (Corrales, 2018).
Constitutional rigidity may also play a role; presidents may be less motivated to try to expand their
powers if the requirements to amend the constitution are high. In support of this, cross- national prec-
edent shows that rigid constitutions experience fewer amendments (Lijphart, 1999).
Research centered on the socioeconomic argument has advanced two main explanations. One
is related to the political regime. Pennings (2003) claimed that established democracies are better
equipped to counteract the executive because they possess stronger, more developed legislatures
than less consolidated democracies. Relatedly, Elkins et al. (2009) showed that stable democra-
cies replace their constitutions less frequently because their leaders are more constrained. Similarly,
Corrales (2018) claimed that regime changes tend to trigger massive constitutional transformations,
offering an opportunity to change presidential powers.
A second explanation is centered on economic factors. Elkins et al. (2009) found that econom-
ically developed countries enjoy more constitutional stability— as long as the constitutions were
written in democratic regimes. The reasoning is that stable constitutions provide the predictabil-
ity necessary to stimulate investment and economic growth (see also Persson & Tabellini, 2005).
Following these authors, presidents of more advanced countries may be more hesitant to try to in-
crease their powers due to the economic uncertainties that such constitutional change can produce in
countries with stable institutions.
Other authors have centered on economic hardship. Przeworski (1991) argued that “shock ther-
apy” reforms lean public opinion in favor of strong presidents. Similarly, Corrales (2018) claimed
that Latin American incumbents have tended to demand more powers when facing economic ad-
versity and that these demands often enjoy popular support. However, Frye (1997) found that the
expansion of presidential powers took place in countries that adopted gradual, not radical, economic
In any case, the causal relationship between socioeconomic factors and presidential powers
remains unclear. Armingeon and Careja (2008) did not find any relation between levels of democ-
racy, GDP per capita, the share of the rural population, telephone penetration, the percentage of
employment in services, and Gini coefficient to the level of executive powers when examining 27
postcommunist countries from 1990 to 2002.
A Psychological Explanation
Between June 2011 and May 2012, I conducted semistructured interviews with 21 former pres-
idents from eight countries. The goal was to start exploring how and why presidential behavior has
an impact on executive governance. Presidents were asked questions related to their relationship
with the constitution, their individual differences and trajectory, the relative importance of the char-
acteristics of presidents for executive governance, and the political context in which they governed
(questions are in the appendix in the online supporting information).
Without exception, the interviewees endorsed the proposition that the individual differences of
presidents have palpable outcomes on executive politics. The leaders also revealed themselves to be
fully aware of how charters framed their behavior: “Possibly the main constraint [for presidential
behavior] is to exercise the mandate respecting the constitution and laws of the republic,” said Óscar
Arias of Costa Rica (personal communication, August 9, 2011).
Presidents strongly agreed in considering the charter a highly malleable document. Most of the
interviewees had an opinion of how to improve the constitution, some recognized thinking about
changing it, and only a minority said that they felt the charter did not need any changes. Presidents
who preferred to change the constitution but did not try often justified their inaction due to the ex-
pected backlash from other actors. For example, Abel Pacheco of Costa Rica said that he “would
have loved to change the constitution” but “did not try it because how can I change anything with the
support of only six deputies?” (personal communication, August 10, 2011).
Pacheco’s remark supports the idea that the distribution of power constrains presidential behav-
ior. However, most leaders who tried to increase their powers did it in adverse institutional environ-
ments. The president’s party did not enjoy a majority in the Lower House in 16 of the 27 attempts
that occurred from 1945 to 2012. Furthermore, 16 of the 25 overreaching leaders faced a rigid con-
stitution (i.e., that demanded the support of at least three- fifths of the legislature to be amended). On
the other hand, only three of the 52 presidents whose parties controlled the Lower House and did not
face a rigid charter tried to increase their powers.
I argue that the presidents’ attempts to expand their powers cannot be completely explained
by the institutional distribution of power and the broader socioeconomic context because they are
partially rooted in the personality traits of the leaders. That explains why only a minority of leaders
attempt to increase their powers even though all of them could benefit from doing so, and why sev-
eral heads of government tried to reform the constitution even in adverse circumstances.
Preceding research on the expansion of presidential powers has overlooked a well- established
tradition that has used a variety of at- a- distance methods to examine the consequences of the person-
ality traits of presidents on executive governance. These works have focused on different traits of the
leaders, such as their Big Five personality traits (Rubenzer & Faschingbauer, 2004; [Arana Araya,
2021a; Arana Araya and Guerrero Valencia, 2020]), their power, achievement, and affiliation motives
(Winter, 2002), their leadership characteristics (Hermann, 2003), and their intellectual brilliance
Building on this tradition, I propose that risk taking should strongly explain why certain lead-
ers attempt to expand their powers. Risk is a multidimensional concept studied in mathematics, the
natural sciences, social sciences, and the humanities. The reason is simple: “Risks are everywhere…
They arise from our own acts and are imposed on us” (Fischhoff & Kadvany, 2011, p. 1). Different
disciplines emphasize distinct aspects of risk such as its sources, measurement, and consequences.
However, at their core, “risks involve threats to outcomes that we value” (Fischhoff & Kadvany,
2011, p. 22).
The Quest for Uncontested Power
Following Holton (2004), risk is composed of two basic aspects. One is the uncertainty of an
outcome pursued, and the other is that the outcome sought should entail gains. The position of pres-
ident entails numerous uncertainties, including electoral outcomes and pressing domestic or interna-
tional events that must be dealt with. Presidents also regularly make decisions whose consequences
they hardly control, from proposing legislation to issuing decrees. However, presidents who try to
expand their powers consciously engage in a long and complex process with particularly high un-
certainties. The leaders often need to negotiate extensively with Congress and make concessions
to change the constitution. They can also face the opposition of the judiciary, voters, the media,
civic organizations, and national elites. If presidents fail in their attempts, they may suffer extremely
taxing consequences. As mentioned earlier, four leaders were ousted and one was forced to resign
after attempting self- serving constitutional reforms. But even if presidents do not lose office after an
unsuccessful attempt, their political capital can experience significant damage (e.g., President Sixto
Durán- Ballén of Ecuador).
Simultaneously, leaders who expand their powers can significantly consolidate their authority.
The 1999 Venezuelan constitution, for example, allowed President Hugo Chávez to concentrate so
much power that “the costs of being in the opposition, and the obstacles for escaping this status, had
never been higher in Venezuela’s democratic history” (Corrales, 2018, p. 127). Among other things,
Chávez gained the right to be reelected, dissolve the National Assembly, and use referenda to revoke
legislation and change the constitution. Chávez’s preferences were also represented in the expansion
of his term from five to six years, the elimination of the Senate, and the increased state’s control over
the oil sector.
Specialists in risk distinguish between general and domain- specific risk taking, such as career,
financial, health, and social risks (e.g., Dohmen et al., 2011). Here, I allude to a general tendency to
take risks, given that presidents who fail in their attempts may experience losses in multiple domains.
In sum, I hypothesize that:
H1: Presidents with higher levels of risk taking are more likely to try to increase their powers.
Presidents who try to increase their powers also share other characteristics. Previous research
has tried to capture the presidents’ motives to understand the goals for which they mobilize and direct
their skills and resources (Winter, 2002). In line with this research agenda, I propose that overreach-
ing leaders have a strong motivation to consolidate their authority.
To be sure, politicians often reach the presidency after a long- term commitment to acquire
power. For example, Óscar Arias and Miguel Ángel Rodríguez of Costa Rica (personal communi-
cation, August 12, 2011), Eduardo Frei of Chile (personal communication, May 3, 2012), Armando
Calderón Sol of El Salvador (personal communication, July 25, 2011), and Abdalá Bucaram of
Ecuador (personal communication, August 16, 2011) expressed that they were already engaged in
politics as minors. But expanding presidential powers demands a superior level of motivation. To
succeed in their attempts, presidents need to add additional work to their busy agendas and spend
months or years negotiating with multiple actors. For example, Bolivian President Evo Morales con-
voked a constituent assembly to replace the constitution in his first year in power (2006), and he was
deeply involved in the constitution- making process that ended up with a new charter that expanded
his powers in 2009 [Arana Araya, 2016a].
Overreaching presidents are also good at convincing others. The presidential attempts are clearly
self- interested. Therefore, leaders need to display exceptional negotiating skills to persuade legisla-
tors, party elites, voters, and interest groups that stretching the influence of their office is beneficial.
For example, Carlos Menem of Argentina needed the endorsement of the opposition party Unión
Cívica Radical (UCR) to attain the two- thirds of congressional support needed to reform the consti-
tution. To increase his leverage, he called for a nonbinding referendum for the (somewhat popular)
reforms, increasing the pressure on the UCR, and at the same time offered concessions to the UCR
(Lucardi & Almaraz, 2017). This strategy divided the UCR and Menem ultimately succeeded in
Finally, presidents who try to increase their powers particularly enjoy concentrating decision-
making to themselves. When asked about the importance of being “strong leaders,” the 21 inter-
viewees provided two main types of answers that revealed varying levels of support to centralizing
authority. Some leaders— such as Antonio Saca of El Salvador (personal communication, July 26,
2011), Rafael Calderón of Costa Rica (personal communication, August 11, 2011), Nicolás Barletta
of Panama (personal communication, August 17, 2011), and Vinicio Cerezo of Guatemala (personal
communication, July 20, 2011)— highlighted the importance of developing fluid relations with
Congress. However, others— Rafael Callejas of Honduras (personal communication, July 28, 2011),
Abel Pacheco, and Óscar Arias— showed an inclination to amass decision- making through tougher
leadership, including retaliating against rebel progovernment legislators to implement their agendas.
“I would be a liar if I were to say that presidents do not retaliate…Human beings are like that. You
step on my foot and I step on yours,” said Pacheco.
The characteristics mentioned— high motivation, superb negotiation skills, and a strong drive
for concentrating power— are well captured by assertiveness. A study defines assertiveness as “a
person’s tendency to actively defend, pursue, and speak out for his or her own interests” (Ames &
Flynn, 2007, p. 307). High assertiveness has been associated with leadership emergence and effec-
tiveness (Bass & Stogdill, 1990), and with traits such as dominance, aggressiveness, nondeference,
and competitiveness (Ames & Flynn, 2007). Research suggests that assertive individuals are more
willing to engage in conflict (Graziano, Jensen- Campbell, & Hair, 1996), often attain their goals, and
put themselves in advantageous positions in social networks (Ames & Flynn, 2007).
This personality trait is multifactorial, so different measures may capture diverse dimensions
of the construct. To avoid ambiguities, I follow the International Personality Item Pool operational-
ization (IPIP; Goldberg et al., 2006) because the 10 statements used in the scale (shown on Table2)
capture the features of overreaching leaders. The scale identifies the attributes of someone highly
motivated to succeed, that knows how to convince others, and enjoys leading, making decisions, and
taking control of things. Following this discussion, I hypothesize that:
H2: Presidents with higher levels of assertiveness are more likely to try to increase their powers.
Two qualifications seem relevant. The scholarly literature often distinguishes between risk pref-
erences and risk perceptions (e.g., Fischhoff & Kadvany, 2011; Sjöberg, 2000). While risk pref-
erences refer to the individuals’ willingness to take risks and can be measured by behavior, risk
perceptions allude to how people perceive risks. The argument made in this article focuses on risk
preferences and therefore centers on the observable behavior of presidents. It remains agnostic about
presidents’ perceptions, making no assumption about the leaders’ understanding of the risks involved
in expanding presidential powers. The same applies to assertiveness; for this study, it is inconsequen-
tial whether presidents have misconceptions of their influence when trying to expand their powers.
Second, the argument is about a general pattern of behavior and therefore is also indifferent to
how leaders might strategically express their behavioral dispositions. For example, assertive presi-
dents may choose to appear unassertive in certain contexts. However, the expectation is that assertive
presidents will tend to behave differently than unassertive leaders.
An important implication of these qualifications is that the hypotheses do not predict when pres-
idents will try to expand their powers. It is assumed that the leaders will time their attempts when
they perceive a proper context to do so. Noticeably, this implication is different from conventional
arguments. Time is central to the two types of arguments that the literature has proposed to explain
the expansion of presidential powers. Research centered on the institutional distribution of power
The Quest for Uncontested Power
proposes that constitution- makers change the rules based on their expectations of future electoral
results or policy influence, whereas the second research stream expects that changes in the socioeco-
nomic context impact the expansion of presidential powers.
The sample encompasses 152 presidents who governed a Latin American country from 1945
to 2012 for at least six months. The end of World War II is the starting point because it is the onset
of the region’s most democratic period. The dependent variable takes the value of 1 when leaders
attempt to change the constitution to increase their powers and 0 otherwise. Consequently, constitu-
tional reforms that only affected later leaders or that were unrelated to the expansion of powers are
excluded. Presidents are considered to have attempted to increase their powers when they made a
public, recorded attempt of their intention. Although in some cases the leaders pursued other major
constitutional changes (e.g., Hugo Chávez in 1999), the expansion of powers was a central goal in all
the attempts considered. The presidential attempts were identified from biographical data (presented
in the appendix in the online supporting information). The successful attempts identified were cor-
roborated with the respective national constitutions and Negretto’s (2013b) database.
I follow the literature that classifies presidential powers as legislative and nonlegislative (e.g.,
Frye, 1997; Negretto, 2013a; Shugart & Carey, 1992), and use as a reference the only dataset of pow-
ers that covers the entire period (Negretto, 2013b). Legislative powers include the president’s veto
powers, decree authority, exclusive initiative on financial or economic legislation, ability to convoke
Congress for extraordinary sessions, submit urgency bills, issue decrees of legislative content in
emergencies, and submit a bill to referendum. Nonlegislative powers refer to the presidents’ author-
ity over the cabinet, influence in appointments, and congressional or other constraints on leaders (a
complete list is in the appendix in the online supporting information).
Information about the leaders’ personality traits come from an online survey completed by ex-
perts on presidents. The expert survey is the main component of the Presidential Database of the
Americas [Arana Araya, 2017], which also contains biographical data about the heads of government
of the United States and Latin American countries. I conducted an expert survey because it better
avoided the validity problems that may emerge when using other at- a- distance methods that demand
the examination of documents. For example, content analysis and historiometric measurements can
introduce too much error if the information under analysis is excessively limited or unreliable (for a
discussion, see Song & Simonton, 2007). Unfortunately, it was difficult to find reliable biographies
for at least two- thirds of the sample. In contrast, finding experts was feasible because many scholars
who have not published biographies about the leaders yet can be found in places such as universities
and media outlets. Therefore, referring to experts allowed access to the most reliable, extensive, and
I led a research team to identify experts about Latin American leaders. We initially conducted a
search in the WorldCat database using the keywords “presidents” and presidentes (for Spanish and
Portuguese) and also different versions of the names of each leader. Second, we examined publica-
tions about the leaders in Google Scholar and Amazon. Third, we asked nearly 50 professional or-
ganizations that work with historians, political scientists, and journalists to help us identify potential
experts. Finally, we asked survey participants to suggest us the names of other scholars qualified to
assess the leaders.
The survey was sent to the emails of 911 experts. In total, 361 experts from 26 nationalities filled
out 531 online surveys in English, Spanish, or Portuguese. Experts assessed 152 Latin American
leaders. The mean number of evaluators per head of state is three, sufficient for making reliable psy-
chological analyses (Rubenzer & Faschingbauer, 2004, p. 319).
The participants were highly educated (96% completed college and 56% held a Ph.D.) and
knowledgeable about the presidents; 216 questionnaires were answered by evaluators who had met
the leaders at least once. Raters’ average age was 57, 73% of them were men, and most were political
scientists (27%), journalists (19%), and historians (18%). Experts answered standardized psycho-
metric questionnaires designed to measure personality traits, background characteristics of presi-
dents (e.g., socioeconomic origin), and nine items designed to measure their potential biases. These
items were gender, age, nationality, city of residence, educational attainment, sympathy toward the
president, approval of the leader’s performance, number of times they met the chief of state (and if
the contact was professional, friendly, or familiar), and their ideology.
Raters answered scales extensively used in personality research to measure risk taking and as-
sertiveness. I used the Risk- Taking Index (RTI, Nicholson, Soane, Fenton- O’Creevy, & Willman,
2005) shown in Table1. The RTI asks about an individual’s relation to risk in six domains: recre-
ational, health, career, finances, personal safety, and social. While the RTI captures the present and
past behavior of individuals, the survey asked experts to differentiate between the chief executive’s
behavior before reaching office and during their term. This distinction was made because presidents
have incentives to moderate (or hide) their risk propensity once in power, and the leaders’ risk taking
in office may be influenced by unobserved factors that transcend their personality. Two other minor
modifications were introduced. First, the original 5- point scale that went from “never” to “very
often” was simplified to a “yes” (coded as 1) or “no” (coded as 0) question to avoid assuming that
raters have such a detailed knowledge of presidents. Second, the statement “standing for election”
was removed as an example of social risks, given that most leaders in the sample stood for elections.
When an evaluator did not fill out a risk dimension, the score for that item was based on the score
received by the other respondents.
The measure of assertiveness comes from the International Personality Item Pool (IPIP)
(Goldberg et al., 2006), shown in Table2. This 10- statement scale captures the characteristics of
individuals who are highly motivated to succeed, express themselves easily, know how to convince
others, and enjoy leading, taking control of things, and making decisions.
As is conventional in expert surveys (Steenbergen & Marks, 2007), the scores of risk taking and
assertiveness for each leader represent the average value that respondents assigned to them. Scores
were treated as invalid when more than two statements were left unassessed.
Cronbach’s alpha was computed to measure the interrater reliability for the independent vari-
ables, with .83 being the average alpha for Risk Taking and .89 for Assertiveness. These numbers are
above the .70– .80 conventional threshold in psychometrics (Song & Simonton, 2007, p. 315).
A potential concern is that presidents may tend to be both risk taking and assertive. If such is
the case, then both variables would vary little and be highly correlated and therefore have little sep-
arate explanatory power. However, Figure1 shows that presidents vary significantly in risk taking
Table 1. Risk Taking
We are interested in the president’s attitude towards risk. Do any of the following descriptions apply to the president
before her term in office and during her term in office?
Before Term During Term
Recreational risks (e.g., rock- climbing, scuba diving)
Health risks (e.g., smoking, poor diet, high alcohol consumption)
Career risks (e.g., quitting a job without another to go to)
Financial risks (e.g., gambling, risky investments)
Safety risks (e.g., fast driving, city cycling without a helmet)
Social risks (e.g., publicly challenging a rule or decision)
Source: Nicholson et al. (2005).
The Quest for Uncontested Power
and assertiveness, and this holds true irrespective of whether they tried to increase their powers
(triangles) or not (circles). Noticeably, while the leaders in the sample cover the entire risk- taking
spectrum (with 25 fully risk- averse politicians and three full risk takers), their level of assertiveness
ranges from 2.05 (one president) to 4.8 (two leaders) in the 1– 5 scale.
Five control variables are used to address the argument about the institutional distribution of
power. Presidents’ Party is a dichotomous variable that captures whether the president’s party enjoys
a majority in the assembly or not and is taken from Pérez- Liñán, Schmidt, and Vairo (2019). Higher
Courts captures the independence of high courts. It is an index taken from Coppedge et al. (2017)
that ranges from 0 (the courts’ ruling “always” reflect government wishes) to 4 (the court is com-
pletely independent of the government). Amendment captures the requirements needed to change the
constitution. This measure of constitutional rigidity ranges from 0, when there is no Congress or need
of congressional support, to 5, when the president needs more than two- thirds of the legislature to
reform the charter as well as an additional step (e.g., the support from a majority of state legislatures,
as in Mexico). The variable is taken from the national constitutions. Finally, to identify powerful
presidencies, Legislative Powers and NonLegislative Powers take the value of 1 when presidential
powers are one standard deviation above the mean in Negretto’s (2013b) scale, and 0 otherwise.
Two variables address the argument that associates socioeconomic factors to the expansion of
powers. Regime Level measures the regime authority spectrum using the Polity IV 21- point scale.
GDP per Capita (measured in 2000 US$) is used as a proxy for economic development and is taken
from the World Bank (2014).
Since the dependent variable reflects the individual attempts to change the charter (a binary
choice), and because the hypotheses refer to individual predictors, I estimate discrete- time- duration
models in which the unit of analysis is president- year. The analysis is longitudinal to test the hy-
potheses while controlling for arguments relevant in preceding research, in which the timing of
the expansion of presidential powers was relevant. Cubic polynomials are included to control for
temporal dependence and minimize biased estimations when using binary data (Carter & Signorino,
Table 2. Assertiveness
Please indicate the extent to which you agree or disagree with the following characteristics as they apply or not to the
nor Disagree Agree
1. Expressed himself easily
2. Tried to lead others
3. Automatically took charge
4. Knew how to convince others
5. Was the first to act
6. Took control of things
7. Waited for others to lead the way
8. Let others make the decisions
9. Was not highly motivated to succeed
10. Couldn’t come up with new ideas
Note Questions 7– 10 receive the reverse scores of questions 1– 6.
Source: Goldberg et al., 2006.
10 Arana Araya
2010). Therefore, Time in Office (also squared and cubed) captures the number of years presidents
Table3 presents seven probit models. Model 3.1 tests the conventional arguments about the
expansion of presidential powers. Model 3.2 presents the benchmark model— adding Risk Taking
and Assertiveness to 3.1— while 3.3 only considers the successful attempts to examine if the results
change if failed attempts are excluded. Models 3.4– 3.7 address potential biases.
An endogeneity bias could occur if experts assessed personalities based on whether presi-
dents tried to expand their powers or not. If such reverse- coding happened, then Risk Taking and
Assertiveness could be partially explained by the values of the dependent variable. This potential
bias seems unlikely because experts were asked to reflect on the presidents’ characteristics before
they reached office, and because the short statements shown in Tables 1 and 2 are about patterns
of behavior unrelated to constitutional reforms. However, to be conservative, model 3.4 presents a
probit extended- regression model with instrumental variables (IV). The logic is that the potential en-
dogeneity can be corrected if other factors— the IV— are identified such that they are correlated with
Risk Taking and Assertiveness, and they only affect the dependent variable through the endogenous
Research has shown that Risk Taking is correlated with educational attainment (Jung, 2015)
and antisocial behavior (Mishra, Lalumiere, & Williams, 2017), while Assertiveness has been
linked to age (Kimble, Marsh, & Kiska, 1984), being a lawyer (Nevill & Schlecker, 1988), and
religiosity (Bolsinger & McMinn, 1989). Using biographical data, I use these factors as IV. The
exclusion restriction demands that the IV are not directly related to the presidential attempts. To
address this, I tested whether the instruments predict the outcome after controlling for Risk Taking
and Assertiveness. None of the IV were statistically associated with the presidential attempts when
tested individually or jointly (using an F- test). Furthermore, the first- stage model F- statistic is 45.90
for Risk Taking and 30.77 for Assertiveness, above conventional thresholds for instrument strength
(Stock & Yogo, 2005).
Figure 1. Distribution of personality traits among leaders.
The Quest for Uncontested Power
Table 3. Presidential Attempts to Increase Powers
(3.1) (3.2) (3.3) (3.4) (3.5) (3.6) (3.7)
Risk Taking 15.94*** 22.23*** 3.72*** 12.35*** 17.09*** 11.70***
(4.88) (5.74) (0.93) (3.40) (4.58) (3.91)
Assertiveness 5.78*** 8.65*** 1.17*** 7.10*** 6.66*** 4.74***
(1.97) (2.52) (0.45) (2.26) (1.81) (1.62)
President’s Party −0.29 0.91 1.78 −0.05 0.56 0.96 −0.29
(1.15) (1.85) (1.90) (0.17) (2.21) (1.75) (1.61)
Higher Courts −3.02*** −3.85*** −2.42** −0.28** −4.88*** −4.46*** −3.41***
(0.76) (1.13) (1.04) (0.13) (1.12) (0.97) (0.84)
Amendment 0.14 −0.14 0.32 −0.03 −0.25 −0.08 −0.03
(0.33) (0.49) (0.53) (0.05) (0.65) (0.45) (0.41)
Leg. Powers 2.55** 3.90** 5.54*** 0.20 3.89*** 4.50*** 3.41***
(1.01) (1.53) (1.70) (0.17) (1.44) (1.57) (1.33)
Nonleg. Powers 3.80*** 6.62*** 8.34*** 0.24 7.81*** 7.28*** 5.46***
(1.22) (2.42) (2.44) (0.26) (2.12) (2.17) (1.91)
Regime Level 0.05 −0.00 −0.32 0.01 0.01 0.01 0.05
(0.10) (0.17) (0.20) (0.02) (0.18) (0.17) (0.14)
GDP per Capita 0.63* 0.59 0.47 0.03 1.17** 0.52 0.44
(0.34) (0.51) (0.57) (0.05) (0.58) (0.53) (0.42)
Time in Office 0.81 1.60 0.70 0.07 1.48 1.77 1.13
(0.61) (1.86) (1.96) (0.18) (1.05) (1.91) (0.86)
Time in Office2 −0.09 −0.14 0.14 −0.01 −0.17 −0.16 −0.12
(0.11) (0.53) (0.62) (0.05) (0.18) (0.56) (0.13)
Time in Office3 0.00 0.00 −0.02 −0.00 0.00 0.00 0.00
(0.00) (0.04) (0.06) (0.00) (0.00) (0.05) (0.00)
Constant −14.16*** −47.72*** −63.80*** −6.79*** −53.13*** −52.63*** −36.93***
(1.80) (10.84) (12.95) (1.58) (9.43) (9.73) (8.18)
First Stage: Instrumental Variables
I. Movts. 0.06***
Errors Risk −0.72***
Errors Assert. −0.52**
Observations 670 670 672 663 574 670 642
Presidents 122 122 122 121 102 122 116
Note Probit regression with standard errors in parentheses.
12 Arana Araya
Another bias could emerge if some of the experts’ characteristics systematically affected their
assessments. Inspired in Martínez i Coma & van Ham (2015) and Rubenzer and Faschingbauer
(2004), I correlated the experts’ 22 answers with their own personal characteristics. Their sympathy
toward the leaders and their judgment of the presidents’ performance were the only features that
correlated above .20 with one of the scores the leaders received. These two variables have a correla-
tion of .81, arguably because they both measure the raters’ positive/negative inclination toward the
president. Based on this exercise, model 3.5 reruns the benchmark model (3.2) excluding experts
who strongly (dis)approved the leaders’ performances.
Since I did not find experts to assess 150 of the 302 presidents who governed the region from
1945 to 2012, there could be a sample- selection bias whereby only certain leaders were assessed.
To account for a censored selection effect, I follow the Heckman (1976) two- stage procedure and
include the Inverse Mills Ratio (IMR) as an extra explanatory variable in model 3.6 (the first- stage
model is in the appendix in the online supporting information).
Finally, model 3.7 excludes heads of government who may not have had enough support to re-
form the constitution because they were either interim (i.e., they served to complete someone else’s
term) or “puppet” presidents (i.e., they governed under the orders of someone else). The data for
these variables is taken from Mainwaring and Pérez- Liñán (2013).
Noticeably, Risk Taking and Assertiveness are statistically associated with the likelihood that
presidents will try to increase their powers across all models. Model 3.3 shows that the effect size of
the variables is even bigger when only successful attempts (21 out of the 25 attempts) are considered.
Figure2 shows the estimation of predicted probabilities based on model 3.3. The chances that
presidents will increase their powers goes from 0.1% when presidents score zero in Risk Taking (25
leaders) to 29% when they have the maximum score (e.g., Fernando Collor de Mello of Brazil).
Similarly, the probabilities of observing a self- serving constitutional reform moves from almost zero
to 10% when the score in Assertiveness for the heads of government increases from 3 to 4.8 (Eduardo
Haedo and Benito Nardone, both of Uruguay). The effect size when all presidential attempts are
considered (model 3.2) is also substantive: The chances that extreme risk- takers and highly assertive
leaders will try a constitutional reform are 20% and 9%, respectively.
The results for the conventional arguments are mixed. The two socioeconomic factors examined
are not statistically significant, but three of the five variables that capture the institutional distribution
of power attain statistical significance. All models show that presidents are less likely to attempt a
constitutional reform if higher courts are more independent. In all models but 3.4, the leaders are also
more likely to try to expand their authority if they already enjoy strong legislative and nonlegislative
Figure 2. Predicted probabilities of constitutional changes.
The Quest for Uncontested Power
To closely examine whether the benchmark model (3.2) improves the predictive capacity of the
conventional arguments (tested in 3.1), I conducted a Receiver Operator Characteristic (ROC) plot.
The AUC (“Area Under the Curve”) is equal to the probability that the simulation predicts a ran-
domly chosen, positive observed instance of a president attempting to increase his or her powers as
more probable than a randomly chosen, negative instance. The gray curve in Figure3 shows that 3.1
has a predictive capacity of 65%, which increases to 74% when the plot is estimated with 3.2 (under
the black curve). This result suggests that the literature has identified relevant causes of the presi-
dential attempts, but that adding the personality traits further improves our understanding of them.
Model 3.4 reveals that the relationship between the independent variables and the presidential
attempts remains consistent when the two personality traits are treated as endogenous variables. The
instruments used show that presidents who participated in illegal movements and are less educated
are more risk takers, while those who are lawyers, less religious, and younger tend to be more asser-
tive. Table3 also shows that the results hold when raters with strong assessments of the presidents’
performance are excluded (model 3.5), when considering a potential sample- selection bias (model
3.6), and when excluding puppets and interim leaders (model 3.7).
The results are reestimated through seven robustness checks. Model 4.1 includes an alternative
measurement of Risk Taking. I conducted a factor analysis to explore whether there are independent
latent variables in the RTI. The exercise revealed two factors, composed of what can be considered
as private risks (recreational, health, financial, and safety risks) and public risks (career and social
risks). Presidents who attempt to increase their powers may be considered to take public rather than
private risks, so I only included these types of risks to capture Risk Taking.
Models 4.2 and 4.3 examine whether other individual characteristics of the leaders might ex-
plain their attempts. Model 4.2 controls for political ideology. This variable is taken from the expert
Figure 3. Receiver operator curves for models 3.1 (gray) and 3.2 (black).
14 Arana Araya
Table 4. Robustness Checks on Presidential Attempts
4.1 4.2 4.3 4.4 4.5 4.6 4.7
Public Risk Ideology Experience Interactions Diffusion
Support GDP Growth
Risk Taking 11.98*** 14.88*** 13.45** 14.48*** 12.48*** 13.97***
(4.18) (4.15) (6.52) (4.27) (3.80) (4.45)
Assertiveness 4.05** 4.62** 5.65*** 6.75** 5.37*** 3.71** 5.24***
(1.77) (1.90) (1.78) (2.69) (1.79) (1.89) (1.69)
New Variable 2.49 −0.33 0.35 0.89 −1.82 5.72
(2.54) (0.56) (0.55) (1.99) (5.00) (13.28)
President’s Party −0.50 −0.14 0.36 0.07 0.71 0.79
(1.44) (1.63) (1.61) (1.95) (1.70) (1.54)
Higher Courts −3.14*** −3.65*** −4.23*** −3.47** −3.56*** −4.91*** −2.88***
(1.20) (0.89) (0.93) (1.40) (0.98) (1.39) (0.94)
Amendment −0.51 −0.12 0.03 −0.09 −0.15 −0.05 −0.05
(0.60) (0.43) (0.42) (0.53) (0.45) (0.64) (0.37)
Leg. Powers 3.46** 3.48** 4.14*** 4.37*** 0.02 2.85* 3.30**
(1.67) (1.45) (1.45) (1.61) (0.16) (1.53) (1.32)
Nonleg. Powers 2.42 5.64*** 6.52*** 6.67** 0.56 6.34** 5.34***
(2.03) (1.94) (2.09) (3.03) (0.45) (2.50) (1.92)
Regime Level 0.07 0.04 0.04 1.89 3.38** 0.19 −0.01
(0.13) (0.15) (0.15) (1.52) (1.38) (0.19) (0.16)
GDP per Capita 0.78* 0.51 0.62 0.44 5.96*** 0.77
(0.47) (0.43) (0.46) (0.69) (2.10) (0.58)
Time in Office 1.18 1.30 1.55 1.23 1.40 0.09 1.11
(0.98) (1.47) (1.80) (1.85) (1.54) (2.18) (1.43)
Time in Office2 −0.15 −0.12 −0.15 −0.11 −0.12 0.45 −0.07
(0.17) (0.40) (0.51) (0.57) (0.44) (0.92) (0.40)
Time in Office3 0.00 0.00 0.00 −0.00 0.00 −0.08 −0.00
(0.00) (0.03) (0.04) (0.05) (0.04) (0.11) (0.03)
Risk * Regime 0.01
Assert. * Regime −0.49
Constant −28.48*** −36.05*** −47.37*** −46.98*** −43.46*** −35.30*** −39.91***
(9.47) (10.81) (9.41) (13.25) (9.34) (9.22) (8.37)
Observations 614 670 670 670 670 621 679
Presidents 109 122 122 122 122 117 123
Note Probit regression with standard errors in parentheses.
The Quest for Uncontested Power
survey. Participants were asked to place the political ideology of leaders in a scale that goes from
1 (far left) to 7 (far right). Model 4.3 controls for political experience. This scale, taken from the
leaders’ biographies, is based on the positions held by the leaders before reaching office. It takes the
value of 1 when presidents did not have political experience; 2 if they held a position at the subna-
tional level or lead a party or had a low position in the executive branch; 3 if the head of government
was a legislator or minister or vice- minister; and 4 if the leader was a legislator and minister or vice-
minister, or was previously president.
Given that leaders may have more leeway to express their behavioral dispositions when they face
less institutional constraints (i.e., in less democratic environments), model 4.4 interacts Assertiveness
and Risk Taking with Regime Level. Model 4.5 tests for Diffusion effects, capturing whether leaders
are more likely to try to expand their powers when a leader from a neighboring country attempted to
do so in the previous year. Models 4.6 and 4.7 reexamine conventional arguments. Model 4.6 reex-
amines the institutional argument by including the proportion of the Lower House that supports the
head of government (taken from Pérez- Liñán et al., 2019). Model 4.7 reexamines the socioeconomic
argument by including GDP Growth. The data is taken from the World Bank (2014) for years since
1960 and from Mainwaring and Pérez- Liñán (2013) for earlier years.
Table4 shows that the personality traits remain statistically significant across all models except
4.1, which reveals that using a narrower version of risk taking loses explanatory power. Models 4.2
and 4.3 show that the ideology and the political experience of the leaders are unrelated to their at-
tempts to increase their powers. According to model 4.4, the effect of Risk Taking and Assertiveness
on the presidential attempts to increase powers is not conditional on Regime Level. Model 4.5 sug-
gests that the leaders do not make their attempts based on whether heads of state from neighboring
countries recently tried to expand their powers. Models 4.6 and 4.7 are consistent with the results
shown on Table3, showing no relationship between the attempts and either the legislative support
that presidents have (4.6) or GDP Growth (4.7).
TableS1 (in the online supporting information) shows more robustness checks. A concern could
be that given that Risk Taking and Assertiveness are time- invariant, a cross- sectional analysis would
be more accurate than a longitudinal study to test the hypotheses (i.e., the unit of analysis should be
president and not president- year). A downside of such an approach is that it impedes incorporating
the time- variant arguments made by preceding research. Furthermore, small samples are charac-
terized by low statistical power, higher false discovery rate, and inflated effect- size estimations.
Interestingly, the results hold in the cross- sectional analysis presented on model 5.2, although the
personality traits lose some statistical power. The results on TableS1 also confirm that judicial con-
straints on executive power inhibit the presidential attempts, and that socioeconomic indicators (in-
flation and the dependence of oil and mineral exports) are unrelated to overreaching presidential
This article showed that the personality traits of presidents are an important driver of their at-
tempts to increase their powers. The analysis showed that risk- averse and unassertive Latin American
presidents are almost fully unlikely to try to expand their powers, but the chances that highly risk-
taking and assertive leaders will try to reform the constitution (and succeed) are much greater. The
ROC plot showed a significant increase in the predictive capacity of the model that captured con-
ventional arguments when the personality traits were added. While more studies are needed to fully
understand the potential consequences of having risk- taking and assertive presidents, the results
suggest that leaders with these attributes may lead autocratization processes through self- serving
constitutional reforms. In doing so, this article has shed light on characteristics that overreaching
presidents have and on the processes by which they consolidate their power.
16 Arana Araya
This study makes a novel contribution by simultaneously testing presidency- oriented and
president- centered research on a cross- national sample of Latin American presidents. Presidency-
oriented research attributes policy outcomes to the institutional setting in which the leaders govern
(e.g., Lewis, 2008) and often treats the uniqueness of leaders as unimportant and incomparable. In
contrast, president- centered works try to understand executive politics through the analysis of the
personal characteristics of the presidents (e.g., [Arana Araya, 2016b]). Both research streams have
often existed in parallel lanes, leading to contradictory views on how the presidency works.
The results also extend preceding research that has incorporated insights from differential
psychology to understand the difference that individuals make in the presidency (Hermann, 2003;
Rubenzer & Faschingbauer, 2004; Simonton, 1986; Winter, 2002). Despite the variety of techniques
and samples covered, within this corpus of research there are often coherent results [see also Arana
Araya, 2021b]. For example, Simonton’s (1986) measure of “forcefulness” for American presidents
resembles the scale used here for assertiveness. In line with this study, the author found that force-
fulness is negatively associated with moderation and positively linked to “achievement drive” and
legislative victories. The accumulated evidence that who the president is affects executive gover-
nance and the consistency across at- a- distance studies questions presidency- oriented research that
depersonalizes the presidency.
Further research could explore whether other individual differences or using other at- a- distance
methods could explain overreaching presidential behavior. Although this work was confined to Latin
American presidents, a psychological approach can also be used to study other political elites (see
also [Arana Araya, 2018]). For example, researchers may explore whether prime ministers with
certain personality traits are more likely to lead the “presidentialization” of European parliamentary
systems, while Africanists may examine if there is any relationship between the characteristics of
rulers and state failure.
The author thanks Daniel Hansen, Aníbal Pérez- Liñán, Alison Munden, Daniel Silverman, Kyle
Robertson, Daniel K. Nedal, John Chin, and four anonymous reviewers for their thoughtful com-
ments and suggestions to improve this article. Correspondence concerning this article should be
addressed to Ignacio Arana Araya, Institute for Politics and Strategy, Carnegie Mellon University,
117 Dunlap St, Pittsburgh, PA 15214, USA. E- mail: firstname.lastname@example.org.
Ames, D. R., & Flynn, F. J. (2007). What breaks a leader: The curvilinear relation between assertiveness and leadership.
Journal of Personality and Social Psychology, 92, 307– 324.
Arana Araya, I. (2021a). The ‘Big Five’ personality traits of presidents and the relaxation of term limits in Latin America.
Democratization, https://doi.org/10.1080/13510 347.2021.1957838
Arana Araya, I. (2021b). The personalities of presidents as independent variables. Political Psychology, 42(4), 695– 712.
Arana Araya, I. (2016a). What drives evo’s attempts to remain in power? A psychological explanation. Bolivian Studies
Journal/Revista de Estudios Bolivianos, 22, 191– 219.
Arana Araya, I. (2016b). ¿Cómo evaluar a los integrantes de la élite política? Una propuesta basada en los presidentes amer-
icanos. Política, 54(1), 219– 254.
Arana Araya, I. (2017). Presidential database of the Americas [v.2]. Unpublished Data Set, Carnegie Mellon University.
Arana Araya, I. (2018). Comparative political elites. In A. Farazmand (Ed.), Global encyclopedia of public administration,
public policy, and governance (pp. 923– 930). Heidelberg, Germany: Springer International Publishing.
Arana Araya, I., & Guerrero Valencia, C. (2020). Executive- legislative relations: When do legislators trust the president? InM.
Alcántara Sáez, M. García Moreno, & C. Rivas Pérez (Eds.), Politics and political elites in Latin America (pp.151– 171).
Cham, Switzerland: Springer.
The Quest for Uncontested Power
Armingeon, K., & Careja, R. (2008). Institutional change and stability in postcommunist countries, 1990– 2002. European
Journal of Political Research, 47, 436– 466.
Bass, B. M., & Stogdill, R. M. (1990). Bass and Stogdill’s handbook of leadership: Theory, research, and managerial appli-
cations. New York, NY: Free Press.
Bolsinger, S. A., & McMinn, M. R. (1989). Assertiveness training and Christian values. Counseling and Values, 34, 21– 32.
Carter, D. B., & Signorino, C. S. (2010). Back to the future: Modeling time dependence in binary data. Political Analysis,
18(3), 271– 292.
Coppedge, M., Gerring, J., Lindberg, S. I., Skaaning, S. E., Teorell, J., Altman, D., … Knutsen, C. H. (2017). V- Dem Dataset
v7.1. Varieties of Democracy (V- Dem) Project.
Corrales, J. (2018). Fixing democracy: How power asymmetries help explain presidential powers in new constitutions, evi-
dence from Latin America. Oxford, United Kingdom: Oxford University Press.
Dohmen, T., Falk, A., Huffman, D., Sunde, U., Schupp, J., & Wagner, G. G. (2011). Individual risk attitudes: Measurement,
determinants, and behavioral consequences. Journal of the European Economic Association, 9, 522– 550.
Elkins, Z., Ginsburg, T., & Melton, J. (2009). The endurance of national constitutions. New York, NY: Cambridge University
Fischhoff, B., & Kadvany, J. (2011). Risk: A very short introduction. New York, NY: Oxford University Press.
Frye, T. (1997). A politics of institutional choice: Post- communist presidencies. Comparative Political Studies, 30, 523– 552.
Goldberg, L. R., Johnson, J. A., Eber, H. W., Hogan, R., Ashton, M. C., Cloninger, C. R., & Gough, H. G. (2006). The interna-
tional personality item pool and the future of public- domain personality measures. Journal of Research in Personality,
40, 84– 96.
Graziano, W. G., Jensen- Campbell, L. A., & Hair, E. C. (1996). Perceiving interpersonal conflict and 3 reacting to it: The case
for agreeableness. Journal of Personality and Social Psychology, 70, 820– 835.
Haggard, S., & Kaufman, R. R. (1994). The challenges of consolidation. Journal of Democracy, 5(4), 5– 16.
Heckman, J. J. (1976). The common structure of statistical models of truncation, sample selection and limited dependent
variables and a simple estimator for such models. Annals of Economic and Social Measurement, 5, 475– 492.
Helmke, G. (2017). Institutions on the edge: The origins and consequences of inter- branch crises in Latin America. Cambridge,
United Kingdom: Cambridge University Press.
Hermann, M. G. (2003). Assessing leadership style: Trait analysis. In J. M. Post (Ed.), The psychological assessment of polit-
ical leaders: With profiles of Saddam Hussein and Bill Clinton (pp. 178– 212). Ann Arbor, MI: University of Michigan
Holton, G. A. (2004). Defining risk. Financial Analysts Journal, 60, 19– 25.
Jung, S. (2015). Does education affect risk aversion? Evidence from the British education reform. Applied Economics, 47(28),
Kimble, C. E., Marsh, N. B., & Kiska, A. C. (1984). Sex, age, and cultural differences in self- reported assertiveness.
Psychological Reports, 55, 419– 422.
Larkins, C. (1998). The judiciary and delegative democracy in Argentina. Comparative Politics, 30, 423– 442.
Levitsky, S., & Ziblatt, D. (2018). How democracies die. New York, NY: Crown.
Lewis, D. E. (2008). The politics of presidential appointments: Political control and bureaucratic performance. Princeton,
NJ: Princeton University Press.
Lijphart, A. (1999). Patterns of democracy: Government forms and performance in thirty- six countries. New Haven, CT: Yale
Linz, J. J. (1990). The perils of presidentialism. Journal of Democracy, 1, 51– 69.
Lucardi, A., & Almaraz, M. G. (2017). With a little help from the opposition? Relaxing term limits in the argentine provinces,
1983– 2017. Journal of Politics in Latin America, 9(3), 49– 90.
Lührmann, A., & Lindberg, S. I. (2019). A third wave of autocratization is here: What is new about it? Democratization, 26(7),
Mainwaring, S., & Pérez- Liñán, A. (2013). Democracies and dictatorships in Latin America: Emergence, survival, and fall.
Cambridge, United Kingdom: Cambridge University Press.
Mainwaring, S., & Shugart, M. (Eds.). (1997). Presidentialism and democracy in Latin America. Cambridge, United Kingdom:
Cambridge University Press.
Marshall, M. G., Jaggers, K., & Gurr, T. R. (2014). Polity IV Annual Time- Series, 1800– 2013. Center for International
Development and Conflict Management at the University of Maryland College Park. Retrieved from http://www.syste
micpe ace.org/polit y/polit y4x.htm
Martínez i Coma, F., & van Ham, C. (2015). Can experts judge elections? Testing the validity of expert judgments for mea-
suring election integrity. European Journal of Political Research, 54, 305– 325.
18 Arana Araya
Mishra, S., Lalumiere, M. L., & Williams, R. J. (2017). Gambling, risk- taking, and antisocial behavior: A replication study
supporting the generality of deviance. Journal of Gambling Studies, 33(1), 15– 36.
Negretto, G. (2009). Political parties and institutional design: Explaining constitutional choice in Latin America. British
Journal of Political Science, 39, 117– 139.
Negretto, G. (2013a). Making constitutions: Presidents, parties, and institutional choice in Latin America. Cambridge, United
Kingdom: Cambridge University Press.
Negretto, G. (2013b). Latin American constitutional change database. Retrieved from http://www.la- const ituti onalc hange.
Nevill, D. D., & Schlecker, D. I. (1988). The relation of self- efficacy and assertiveness to willingness to engage in traditional/
nontraditional career activities. Psychology of Women Quarterly, 12, 91– 98.
Nicholson, N., Soane, E., Fenton- O’Creevy, M., & Willman, P. (2005). Personality and domain- specific risk taking. Journal
of Risk Research, 8, 157– 176.
Pennings, P. (2003). Beyond dichotomous explanations: Explaining constitutional control of the executive with fuzzy- sets.
European Journal of Political Research, 42, 541– 567.
Pérez- Liñán, A., Schmidt, N., & Vairo, D. (2019). Presidential hegemony and democratic backsliding in Latin America, 1925–
2016. Democratization, 26(4), 606– 625.
Persson, T., & Tabellini, G. E. (2005). The economic effects of constitutions. Cambridge, MA: MIT Press.
Przeworski, A. (1991). Democracy and the market: Political and economic reforms in Eastern Europe and Latin America.
Cambridge, United Kingdom: Cambridge University Press.
Rubenzer, S. J., & Faschingbauer, T. R. (2004). Personality, character, and leadership in the White House: Psychologists
assess the presidents. Washington, DC: Bassey’s.
Runciman, D. (2018). How democracy ends. London, United Kingdom: Profile Books.
Shugart, M. S. (1993). Of presidents and parliaments. East European Constitutional Review, 2, 30– 32.
Shugart, M. S., & Carey, J. M. (1992). Presidents and assemblies: Constitutional design and electoral dynamics. Cambridge,
United Kingdom: Cambridge University Press.
Simonton, D. K. (1986). Presidential personality: Biographical use of the Gough Adjective Check List. Journal of Personality
and Social Psychology, 51(1), 149– 160.
Sjöberg, L. (2000). Factors in risk perception. Risk Analysis, 20, 1– 12.
Song, A. V., & Simonton, D. K. (2007). Personality assessment at a distance. In R. W. Robins, R. C. Fraley, & R. F. Krueger
(Eds.), Handbook of research methods in personality psychology (pp. 308– 201). New York, NY: Guilford.
Steenbergen, M. R., & Marks, G. (2007). Evaluating expert judgments. European Journal of Political Research, 46, 347– 366.
Stock, J. H., & Yogo, M. (2005). Testing for weak instruments in linear IV regression. In D. W. K. Andrews & J. H. Stock
(Eds.), Identification and inference for econometric models: Essays in honor of Thomas Rothenberg (pp. 80– 108). New
York, NY: Cambridge University Press.
Winter, D. G. (2002). Motivation and political leadership. In L. O. Valenty & O. Feldman (Eds.), Political leadership for the
new century: Personality and behavior among American leaders (pp. 25– 47). Westport, CT: Praeger.
World Bank. (2014). World Development Indicators 1960– 2014. Retrieved from http://data.world bank.org/data- catal og/world
- devel opmen t- indic ators
Additional supporting information may be found in the online version of this article at the publisher’s
Table S1. Additional Robustness Checks on Presidential Attempts to Increase Powers
Figure S1. Predicted probabilities of constitutional changes.
Table S2. Attempts at Constitutional Change
Table S3. Presidential Legislative Powers
Table S4. Presidential Non- Legislative Powers
Table S5. Probability that a President will be Assessed
Table S6. Interviews
Table S7. Questionnaire Applied to Presidents