Access the full text.
Sign up today, get DeepDyve free for 14 days.
Loading next page...
/lp/wiley/retaining-political-talent-a-candidate-centered-theory-of-primary-rIBb2509aS
- Publisher
- Wiley
- Copyright
- © 2023 by the Midwest Political Science Association.
- ISSN
- 0092-5853
- eISSN
- 1540-5907
- DOI
- 10.1111/ajps.12776
- Publisher site
- See Article on Publisher Site
Abstract
Verification MaterialsThe data and materials required to verify the computational reproducibility of the results, procedures, and analyses in this article are available on the American Journal of Political Science Dataverse within the Harvard Dataverse Network, at: https://doi.org/10.7910/DVN/GRVR99.In July 2010, Mario López Valdez, popularly known as “Malova,” won a groundbreaking gubernatorial election that ended over eight decades of uninterrupted rule by the Partido Revolucionario Institucional (PRI) in the Mexican state of Sinaloa. This outcome came as a surprise for many observers of Mexican politics. Just six months prior, the real struggle for power was taking place within the PRI itself, with Malova, then a member of that party, trying to win the party's nomination over Jesús Vizcarra, the sitting governor's choice to succeed him in office. By late March, when it became clear that Malova would not become the PRI's candidate, he left the party and secured the nomination of a pre‐electoral coalition formed by the conservative Partido Acción Nacional (PAN) and the leftist Partido de la Revolución Democrática (PRD). Although the circumstances surrounding this election might seem unusual, cases like this are common in Mexican politics (Barrow 2007; Kerevel 2014).1Party defections due to party leaders blocking career paths are also common in other regions, including sub‐Saharan Africa, South and Southeast Asia, and Eastern Europe (see Fashagba 2014, Fell 2013, Shabad and Slomczynski 2004, Zhirnov and Mufti 2019).This Mexican election illustrates a central feature of electoral and legislative politics in many developing countries and emerging democracies—the weakness of politicians' attachments towards political parties (Desposato 2006; Hicken 2006; Shabad and Slomczynski 2004). The fragility of party loyalties poses formidable challenges for political parties.2Notably, party switching and high electoral volatility, both of which are linked to weak party institutionalization (Mainwaring and Scully 1995). Even in settings where partisan attachments are strong, politicians often face incentives to pursue their personal interests at the expense of their parties' collective goals. Indeed, an extensive literature has developed around the role of parties in creating institutional arrangements that commit their members to act in the party's best interest (Aldrich 1995; Cox and McCubbins 1993). However, as politicians attach less value to their party labels, attaining this goal becomes increasingly difficult. How do parties manage to achieve their electoral goals under these conditions? How can parties present a united front against their rivals when many of their members behave as free agents?This article argues that in contexts in which the linkages between politicians and parties are weak, party leaders can keep their parties united by choosing institutions that limit their own influence in candidate selection. I formalize this argument by constructing a model of intraparty politics and electoral competition. The model describes the interaction between a leader and an aspirant to office within a political party; the leader's choice of a nomination rule for the party is followed by the aspirant's decision of whether to stay in the party or switch parties. After this stage, each party nominates a general election candidate from its pool of aspirants, and then an election takes place. In the model, intraparty politics and electoral competition are connected through the office‐seeker's party‐affiliation decision, which, due to the lack of a meaningful attachment to any one party, is based exclusively on office‐seeking considerations.I characterize an equilibrium in which the party leader is strictly better off by relinquishing control over the party's nomination and adopting primary elections. Nomination rules that allow leaders to handpick the party's candidates create a commitment problem that can trigger costly defections from high‐quality aspirants. By adopting primaries, party leaders can overcome this commitment problem, and thus, retain top political talent within the party's ranks. This mechanism is substantively different from those advanced in other formal accounts of the adoption of primaries, which typically emphasize the leaders' uncertainty over the quality of the aspirants seeking the nomination (e.g., Serra 2011). Indeed, a central feature of the model is that primaries are used in equilibrium even though the leader knows exactly who is the most electable politician in their party and, if they wanted to, could simply handpick this aspirant.The theoretical model suggests two empirical implications. First, the party that adopts primaries must be electorally strong ex ante. This occurs because an aspirant with a credible exit option would only stay in a party that makes them compete for the party's ticket if their chances of winning the general election upon winning the primary are sufficiently high. Second, the top two aspirants within the party must be of similar quality. This reflects the fact that defections are costlier for the party leader as the relative quality of the potential defector increases. Thus, the leader retains control over candidate selection even if it triggers defections from weak aspirants, but adopts primaries to prevent high‐quality aspirants from leaving the party.I provide support for the theory using two data sets on candidate selection in Mexico, a country in which the assumption of weak party‐politician linkages holds well (Barrow 2007; Kerevel 2014). First, I examine the nomination processes of two of the country's major parties, the PAN and PRD, in three legislative elections. In each of these elections, a subset of these parties' candidates were appointed by their party leaders, while the rest were selected in primaries. This allows me to analyze the party leaders' choices while holding constant any party‐ and election‐specific factors. Consistent with the theory, I find parties are more likely to use primaries where they are electorally stronger ex ante. I present further evidence, including data on candidate entry in primaries, that supports the model's expectations.Second, I analyze gubernatorial candidate selection within the PRI during the last years of the party's hegemonic rule. Traditionally, the president would handpick the PRI's candidates but, as the country transitioned to democracy in the late 1990s, the party experimented with primaries in several states (Díaz‐Cayeros and Langston 2004). While this trend has been taken as evidence that parties expecting a weak electoral performance have greater incentive to adopt primaries, my analysis—which exploits the variation in nomination rules across states—reveals a different picture. In line with the model, I find the PRI adopted primaries where (1) it expected better electoral results, and (2) the party's top gubernatorial hopefuls were of similar quality.The theory advanced in this article has implications beyond the Mexican case. Over recent years, several parties in developing countries and new democracies have voluntarily used primaries to nominate their candidates for office. Out of 78 presidential elections held in Latin America during 1978–2004, there were 30 in which at least one party voluntarily selected its candidate in a primary (Kemahlioglu, Weitz‐Shapiro, and Hirano 2009). Evidence from sub‐Saharan Africa reveals a similar pattern, with 15 countries in which at least one major party has used primaries to nominate its legislative candidates in the last decade (Ichino and Nathan 2018).This article is part of a growing literature regarding the voluntary adoption of primaries. Several scholars argue that parties choose primaries because they provide an electoral premium. Carey and Polga‐Hecimovich (2006) argue that in Latin America, where primaries remain the exception, voters might reward parties that do use them because they provide a democratic “seal of approval.” Other, mostly formal, accounts identify an electoral bonus that stems from the fact that primaries reveal information about the aspirants' quality, allowing parties to nominate more electable candidates (Adams and Merrill 2008; Serra 2011; Snyder and Ting 2011).A second, mostly empirical, literature views the adoption of primaries as the party leader's response to intraparty strife (e.g., Astudillo and Detterbeck 2020; De Luca, Jones, and Tula 2002). Within this group, there are substantive differences in terms of how intraparty conflict is depicted. While some focus on “horizontal” disputes among factions (Kemahlioglu, Weitz‐Shapiro, and Hirano 2009), others emphasize “vertical” conflict between leaders and activists (Bruhn 2014). Similarly, some depict conflict as an outcome of a party's ideological heterogeneity (Hortala‐Vallve and Mueller 2015) and others as driven by competition for rents (Ichino and Nathan 2012).The theory advanced in this article contains elements of both these approaches. In the model, a primary bonus emerges in equilibrium too. Whereas in existing works this bonus materializes mechanically, the one identified here emerges endogenously from the actors' strategic behavior. Specifically, the primary bonus arises because the leader adopts primaries in order to prevent the defection of a skilled politician, which not only guarantees that their party will have an electable nominee but also shrinks the pool of high‐quality candidates available to their rivals.This article is most closely related to Poiré (2002), who also highlights the role of primaries in preventing defections. In contrast to this author, who focuses on the party leader's calculus to use primaries, I also consider the incentives of the potential defector and study how these actors interact in a game‐theoretic framework. Because of this, I am able to uncover the positive relationship between the use of primaries and the party's ex ante electoral strength. Substantively, this means that only parties that are electorally strong to begin with can strategically adopt primaries to prevent potential defections.This article is also part of a vast literature that studies how electoral institutions shape politicians' behavior (e.g., Hix 2004; Shomer 2017). While legislative politics scholars argue that the use of primaries results in the nomination of disloyal politicians, the theory presented here shows that weak party loyalties can create incentives for the adoption of primaries in the first place. Moreover, the model suggests that whether primaries result in the nomination of disloyal politicians should be conditional on their parties' electoral strength.Finally, this article contributes to the comparative study of party switching. Politicians' party affiliation decisions have been linked to a number of institutional arrangements, such as the electoral system, bicameralism, and federalism (Desposato and Scheiner 2008; Heller and Mershon 2005). Although I do not tackle the puzzle of party switching empirically, the theory suggests that centralized nomination rules can encourage office‐seeking politicians to switch from electorally strong to electorally weak parties.A Candidate‐Centered Theory of Primary AdoptionI develop a model to study the choice of nomination rules within a political party in an environment in which politicians' attachments toward parties are weak. The main goal is to identify the conditions under which party leaders can keep the party united before an election by adopting primaries. The next three subsections lay out the model setup, present the main results, and introduce the model's key implications.SetupConsider an election with two parties, A and B. Before the election, two office‐seekers within Party A struggle over the party's nomination. All other things equal, the leader of Party A (labeled A) prefers one of the aspirants over the other; I refer to this aspirant (labeled I) as the Insider, and to the other (labeled M) as the Maverick.The game begins with A choosing a nomination method m∈{0,1}$m \in \{ 0,1\} $, deciding whether to select the party's candidate in a primary (m=1)$( {m = 1} )$ or to retain the power to handpick the party's nominee (m=0)$( {m = 0} )$. Throughout, I refer to these rules as Primary and Appointment, respectively. Once this rule is in place, the Maverick chooses aM∈{0,1}${a}_M \in \{ 0,1\} $ and decides between staying in Party A(aM=0)$A( {{a}_M = 0} )$ or switching to Party B(aM=1)$B( {{a}_M = 1} )$. If the Maverick stays in Party A, then A's candidate is nominated according to method m. If the rule in place is Appointment, A chooses aA∈{0,1}${a}_A \in \{ 0,1\} $, which accounts to nominating a candidate between the Insider (aA=0)$( {{a}_A = 0} )$ and the Maverick (aA=1)$( {{a}_A = 1} )$. Otherwise, the Insider and the Maverick compete in a primary.If the Maverick switches parties, the Insider automatically becomes Party A's candidate regardless of the nomination method, m. In this instance, Party B's leader (labeled B) chooses aB∈{0,1}${a}_B \in \{ 0,1\} $, deciding whether to nominate the Maverick (aB=1${a}_B = 1$) or an existing party member (labeled D), called the Default candidate (aB=0${a}_B = 0$). Party B only faces this decision when the Maverick joins Party B; when the Maverick stays in Party A, the Default candidate becomes Party B's nominee automatically. Once each party has nominated a candidate, the general election takes place and the game ends.3The Maverick cannot switch parties after losing the primary. Appendix B (pp. 21–22) discusses this modeling decision in detail. Figure 1 shows the game tree and indicates which aspirants run in the general election after each possible sequence of play.1FigureGame Tree of Baseline ModelNotes: Rectangles show pairs of general‐election candidates after each history. There are three possible pairs of general‐election candidates (a,b)$(a, b)$: (Insider, Maverick), (Insider, Default), and (Maverick, Default). The dashed lines denote the primary election: the Insider wins with probability qIM$q_{IM}$ and the Maverick with 1−qIM$1-q_{IM}$. Payoffs are as follows. Parties A and B get a payoff of ω>0$\omega > 0$ when their respective party wins the election. Additionally, I allow parties to ceteris paribus prefer one aspirant over the other. Party A receives payoff ϕA>0${\phi }_A > 0$ when the Insider gets A's nomination. This payoff materializes whether the Insider wins office or not, and it might represent a benefit from strengthening the position of A's faction in intraparty battles or a personal bias based on other factors (such as family ties). Analogously, B gets payoff ϕB>0${\phi }_B > 0$ when the Default candidate gets B's nomination. Finally, the Maverick is exclusively office‐motivated and receives payoff π>0$\pi > 0$ when elected to office, regardless of their partisanship at the end of the game. Consistent with the weak party loyalties assumption, switching parties entails no cost for the Maverick.Next, I specify how the winners of the primary and general election are decided. The main difference between these two types of races is that the outcome of the primary depends exclusively on the relative quality of the aspirants, whereas the general election outcome depends on the relative strength of the parties. To formalize these ideas, each aspirant j∈{I,M,D}$j \in \{ I,M,D\} $ is characterized by a parameter vj>0${v}_j > 0$ that indicates their quality. Here, quality refers to any personal attributes that affect a candidate's ability to mobilize voters, such as their reputation, social connections, or other resource advantages.When the Insider and the Maverick compete in a primary, the outcome is decided according to continuous function f:R→(0,1)$f:\mathbb{R} \to (0,1)$ that maps from the aspirants' relative quality to the probability that the Insider wins the primary. Letting vIM=vI−vM${v}_{IM} = {v}_I - {v}_M$ be the relative quality of these aspirants, the Insider wins with probability f(vIM)$f( {{v}_{IM}} )$. I assume f is strictly increasing, which means the Insider is more (less) likely to win the primary as they become of higher (lower) quality relative to the Maverick. I simplify the notation and say the Insider wins the primary with probability qIM=f(vIM)${q}_{IM} = f({v}_{IM})$, and the Maverick with probability 1−qIM$1 - {q}_{IM}$.4Appendix A (pp. 1–2) considers the relationship between f and different types of primaries.Similarly, let a and b denote the general election candidates of A and B, respectively, and let vab=va−vb${v}_{ab} = {v}_a - {v}_b$ be their relative quality. Because parties are a key determinant of electoral competition, the outcome of the general election depends not only on the quality of the candidates, vab${v}_{ab}$, but also on the parties' relative electoral strength. Substantively, there are multiple sources of a party's electoral strength, including features of the electorate (such as voters' preferences), attributes of the party itself (such as organizational strength), and the interaction of both (such as clientelistic networks). I formalize this notion by letting A win the election with probability f(vab+α)$f({v}_{ab} + \alpha )$, where α captures the electoral strength of A relative to B. I simplify the notation and say A wins the general election with probability pab=f(vab+α)${p}_{ab} = f({v}_{ab} + \alpha )$, and B with probability 1−pab$1 - {p}_{ab}$.I distinguish two general cases depending on whether A is weak or strong relative to B.5This means that whenever A is strong (weak) B is weak (strong). Formally, Party A is strong if α>0$\alpha > 0$. This definition has a very intuitive interpretation; when Party A is strong the chances that A's candidate gets elected are higher than they would be if party labels did not matter. This is so because a wins the general election against b with probability f(vab+α)$f({v}_{ab} + \alpha )$, and thus if parties were not an important determinant of electoral competition this probability would simply be f(vab)$f( {{v}_{ab}} )$. Because f is strictly increasing, when A is strong, a benefits from being Party A's nominee, that is, α>0$\alpha > 0$ implies f(vab)<f(vab+α)$f({v}_{ab}) < f({v}_{ab} + \alpha )$. Intuitively, one can think of α as an additional component of a's quality. Following the same logic, Party A is weak if α<0$\alpha < 0$.The quality of the candidates, the parties' relative strength, and f are public information; that is, all relevant probabilities are known to all the players. Throughout, I treat these probabilities as exogenous. However, Appendix A (p. 1) provides microfoundations to these probabilities by explicitly modeling elections as contests in which candidates compete by spending resources, an approach that accurately captures the type of nonprogrammatic environment considered here.AnalysisThis is a finite extensive game of perfect information, and thus the solution concept I use is subgame perfect equilibrium (referred to as simply equilibrium hereafter). In order to understand the conditions under which using primaries prevents defections, I focus on equilibria in which A holds a contested primary, that is, equilibria in which on the path of play (1) A chooses primary, and (2) the Maverick stays in Party A. The next result provides a set of necessary conditions for such an equilibrium.1PropositionIn equilibrium, Party A selects its candidate in a contested primary only if the following conditions hold:The Insider and Maverick are of similar enough quality, that is, |vI−vM|$| {{v}_I - {v}_M} |$ is not too large.The Maverick is of sufficiently higher quality than the Default candidate, that is, vM−vD${v}_M - {v}_D$ is sufficiently large.Party A is electorally strong, that is, α>0.$\alpha > 0.$The proof (and a more formal statement) are in Appendix A (pp. 4–5). Here, I discuss the intuition behind this result. In this equilibrium, A chooses Primary to overcome a commitment problem created by the Appointment rule. In short, A would like to prevent the Maverick from switching parties by promising to appoint them as the party's candidate. However, if the Maverick were to stay in Party A, forgoing the opportunity to switch parties and become B's candidate, then A is better off nominating the Insider—that is, A cannot credibly promise to handpick the Maverick. Consequently, A choosing Appointment leads the Maverick to switch parties and seek B's nomination. Choosing Primary, then, serves as a commitment device that induces the Maverick to stay in Party A.The commitment problem described above follows from condition (i). This condition allows for two cases: one in which the Maverick is of higher quality than the Insider (vM>vI)$( {{v}_M > {v}_I} )$ and one in which the opposite is true (vI>vM)$( {{v}_I > {v}_M} )$. For expositional purposes only, the following discussion focuses on the case where the Insider is of higher quality than the Maverick, which is the least likely case for the use of primaries. Indeed, existing works explain the choice of primaries as driven by party leader uncertainty over the quality of the aspirant pool, specifically, by the possibility that the aspirant more closely linked to them is of relatively low quality (e.g., Serra 2011). Focusing on the case vI>vM${v}_I > {v}_M$ highlights the fact that the mechanism I propose is not driven by this trade‐off. In fact, a noteworthy feature of the theory is that primaries can emerge in equilibrium even when A knows with certainty that the aspirant most closely linked to them (that is, the Insider) is also the party's most electable aspirant.The equilibrium in Proposition 1 is such that on the path of play (1) A chooses Primary and (2) the Maverick stays in Party A. For this to occur, it must be that off the path of play, the Maverick switches to Party B after A chooses Appointment. Otherwise, since the Maverick would stay in Party A no matter what rule is used, A would be better off choosing Appointment and handpicking the Insider. Therefore, an equilibrium with a contested primary requires that the Maverick commits to switching parties after A chooses Appointment. Moreover, this equilibrium requires that whenever the Maverick switches parties, they get B's nomination. Otherwise, A would know that, no matter what rule they choose, their party's nominee will run in the general election against the Default candidate, and thus the best they can do is choose Appointment and select the Insider. Intuitively, it is clear that the Maverick would only get B's nomination if they are of a higher quality than the Default candidate. Condition (ii) pinpoints how much higher the Maverick's quality has to be for them to secure B's nomination.In this equilibrium, then, Party A's nomination rule completely determines whether the Maverick switches parties. Figure 2 illustrates A's motivation to choose Primary over Appointment. Each panel shows fixed quality levels of the three aspirants, and the arrows indicate which aspirants become the general election candidates. When A chooses Appointment (Panel 2a), the Maverick switches parties and becomes B's nominee. This means the Insider and the Maverick compete in the general election as candidates of A and B, respectively. Alternatively, when A chooses Primary (Panel 2b), the Maverick stays in Party A. Although this generates some uncertainty over the identity of A's candidate, as both the Insider and the Maverick have some positive probability of winning the primary, it guarantees that the primary winner runs in the general election against B's Default candidate.2FigureCandidate Quality and the Choice of Nomination RuleNotes: Figure shows two panels with fixed quality levels for the Insider, the Maverick, and the Default candidate. The figure assumes all conditions in Proposition 1 hold. The figure illustrates that if the Maverick is of high‐enough quality, meaning vM≥v∗$v_M \geq v^{*}$, Party A has incentives to choose Primary. Notice that after choosing Appointment, A wins the general election with probability pIM${p}_{IM}$, which is shaped by the quality of the Insider relative to the Maverick, vI−vM${v}_I - {v}_M$. Therefore, as the quality of the Maverick increases, choosing Appointment becomes less attractive for A. When A chooses Primary the situation is not so clear‐cut. Using primaries is risky for A because the Maverick, who is of lower quality than the Insider, becomes A's candidate with positive probability 1−qIM$1 - {q}_{IM}$. When the Maverick wins the primary, Party A wins the general election with probability pMD${p}_{MD}$, which depends on the quality of the Maverick relative to the Default candidate, vM−vD${v}_M - {v}_D$. Thus, once the Maverick is of high‐enough quality, primaries become less risky for A, since the potential cost of having the Maverick winning the primary diminishes; in the extreme, when the Maverick and the Insider are of equal quality, Party A wins the general election with the same probability regardless of who wins the primary. Overall, if the Maverick's quality exceeds a certain level (v∗${v}^*$ in Figure 2), A has incentives to choose Primary over Appointment in order to retain the Maverick.Finally, the key feature of this equilibrium is that, by choosing Primary, A stops the Maverick from defecting. In the absence of relevant partisan attachments, A's ability to achieve this goal depends exclusively on the Maverick's prospects of winning office in Party A. This is challenging because when the Maverick switches parties, they are guaranteed to run in the general election, but if they stay in Party A they must win the primary in order to compete in the general election. Therefore, for the Maverick to stay in Party A, their chances of winning the general election as A's candidate must be high enough to compensate for the fact that they must compete for the party's ticket. This is what condition (iii) implies. Substantively, while A might have incentives to stop the Maverick from switching parties under a broader set of circumstances, they are only able to do so when A is sufficiently strong.Additional Results I conclude this section by briefly discussing three results and extensions of the model. First, a potential concern with the model is that A chooses its nomination rule but B is assumed to use Appointment. Appendix A (pp. 7–11) presents model extensions that address this asymmetry. Of particular note is one in which A and B simultaneously choose their own nomination rules. In equilibrium, A chooses its candidate in a contested primary under conditions analogous to those in Proposition 1. In short, A has incentives to prevent a defection only if the Maverick's quality is such that B would appoint them as their candidate. If this is indeed the case, the best B can do is choose Appointment in order to make switching more attractive for the Maverick. Thus, for Party A to hold a contested primary in equilibrium it must be that the Maverick stays in Party A even if switching gives them B's nomination with probability one—exactly as in Proposition 1's equilibrium.Second, the model has implications for what the literature calls the primary bonus (Serra 2011). In Appendix A (pp. 5–7), I formally define this concept in the context of the model and identify conditions under which it emerges in equilibrium. The main result shows that only when A is electorally strong ex ante can it increase its chances of winning the general election by selecting its candidate in a primary rather than by an appointment. This result is similar to one found by Snyder and Ting, who argue that “primaries tend to benefit initially favored parties” (Snyder and Ting 2011, 789).Third, I show that primaries cannot always prevent defections. In addition to Proposition 1's equilibrium, there is another type of equilibrium in which A holds an uncontested primary. This equilibrium is trivial in the sense that, as part of the equilibrium strategies, the Maverick switches parties and becomes B's candidate regardless of A's actions. The Maverick's behavior makes A indifferent between both nomination rules, and thus, there is an equilibrium in which A chooses Primary. I highlight two features of this equilibrium: (1) A's primary is uncontested, meaning the Insider runs unopposed and automatically becomes A's nominee, and (2) it emerges when A is not sufficiently strong (see Appendix A, p. 5).6Overall, the model suggests that, when a party uses primaries, there should be a positive association between the party's electoral strength and the presence of disloyal politicians (see Appendix B, pp. 29–30). This is consistent with empirical work on legislative party unity in Mexico (Ascencio and Kerevel 2021).Hypotheses and DiscussionThe model offers two empirical implications about the conditions under which parties voluntarily adopt primaries. I now introduce these hypotheses and discuss their connection to existing work.Hypothesis 1.When party leaders choose to hold primaries, they will do so in areas where their party is electorally strong ex ante.This expectation separates this model from two sets of theories. First, this hypothesis stands in contrast to research arguing that primaries should be adopted by electorally weak parties in order to become more competitive (Adams and Merrill 2008; Serra 2011). Second, and more important, Hypothesis 1 highlights a key difference between the model and other accounts that explicitly link the use of primaries to potential party splits (Hortala‐Vallve and Mueller 2015; Poiré 2002). In contrast to Poiré (2002), who assumes primaries automatically prevent skilled politicians from switching parties, the model shows that using primaries is not sufficient to keep a party united. Indeed, a central insight from Proposition 1 is that only parties that are electorally strong ex ante can prevent costly defections by adopting primaries.Hypothesis 1 is consistent with previous works that document positive associations between parties' electoral strength and their use of primaries (Ichino and Nathan 2012; Kemahlioglu, Weitz‐Shapiro, and Hirano 2009). This pattern has been interpreted as evidence that parties adopt primaries in response to pressure from actors within the party, which are assumed to be more intense as the value of the party's nomination increases. From a theoretical perspective, this interpretation is not fully satisfactory for two reasons. First, it does not explain how these pressures pose a threat to party leaders, that is, whether these actors can credibly commit to creating an electoral penalty for the party. Second, even if party leaders adopt primaries to avoid an electoral penalty, it is unclear why they would do so when the party is electorally dominant and such a penalty would be relatively less costly, rather than when the party expects a close race, and thus, even a small penalty could cost them the election.Intraparty pressures also play an important role in my framework: the Maverick puts pressure on the leaders through the threat of defection. The model studies the conditions under which party leaders will adopt primaries in response to the Maverick's pressures. In this regard, my theory provides two key insights. First, intraparty pressures are necessary but not sufficient for the adoption of primaries. For these pressures to influence the leaders' behavior, they must be backed up by a threat of defection that is both credible and sufficiently costly for the party leaders. Critically, I provide conditions under which the threat of defection will have these properties. Second, only parties that are electorally strong ex ante can prevent defections through the use of primaries. Only if the value of A's nomination is high does the Maverick stay and compete in the primary even though switching parties would result in them getting B's nomination with full certainty. These insights suggest party leaders play a more strategic role than is usually assumed: when the party's ticket is valuable, the leader can take advantage of their party's electoral strength and use primaries in order to retain a high‐quality politician in the party.The second hypothesis states that the party leaders' incentives to adopt primaries are shaped by the relative quality of the aspirants seeking the party's nomination.Hypothesis 2.When party leaders choose to hold primaries, they will do so in areas where the top two aspirants within their party are of similar quality.The intuition for this expectation is illustrated in Figure 3, which shows the outcomes observed in equilibrium for different values of the Maverick's quality, vM${v}_M$. First, a defection from the Maverick is costly only if their quality exceeds level v̲$\underline{{v}} $. When the Maverick is of low enough quality (vM<v̲)$({v}_M < \underline{{v}} )$, A is better off choosing Appointment and selecting the Insider. Thus, only when the Maverick's quality exceeds v̲$\underline{{v}} $ does A have incentive to prevent a defection. When the Maverick's quality is sufficiently high (v¯<vM)$( {\bar{v} < {v}_M} )$, A chooses Appointment and nominates the Maverick. Therefore, A chooses Primary only when the Maverick's quality falls in the interval (v̲,v¯)$(\underline{{v}} ,\bar{v})$. In this range, the Maverick's quality is such that A wants to prevent a defection but at the same time cannot credibly commit to appointing the Maverick. I emphasize that this commitment problem emerges even when the Maverick is of higher quality than the Insider. The reason for this is that the Maverick needs to be high‐enough quality to compensate for the fact that A has an intrinsic preference for the Insider (captured by ϕA${\phi }_A$).3FigureParty A's Equilibrium Behavior as a Function of Maverick's Quality, vM${v}_M$Notes: The panel fixes the Insider's quality, vI$v_I$, and assumes Proposition 1 (conditions (ii) and (iii)) hold. The figure describes Party A's equilibrium behavior as a function of the Maverick's quality, vM$v_M$. This expectation separates the model from theories that link primary adoption to the size of the aspirant pool. Serra (2011) argues that parties are more likely to use primaries when this pool is larger, which increases the likelihood of selecting a candidate who is of higher quality than the one most closely aligned with the party leaders. As discussed, the model shows that the incentive to use primaries is present even when party leaders know with certainty that no other aspirant is of higher quality. Ichino and Nathan (2012) also expect to observe primaries when the nomination is sought by more aspirants but only when party leaders do not have an a priori preferred aspirant.7In their words, “the party leader selects this option [primaries] if he does not have a preferred aspirant … [otherwise] he may consider other options” (Ichino and Nathan 2012, 776). In contrast, I argue primaries emerge because having a preferred aspirant (that is, the Insider) creates a commitment problem for party leaders, which in turn can trigger costly defections. Put differently, I argue party leaders adopt primaries because they cannot credibly promise to not play favorites.Hypothesis 2 is in line with De Luca, Jones, and Tula (2002), who argue that primaries are an attractive option when the national party leader and the local party branch support different aspirants.8However, similar to Hortala‐Vallve and Mueller (2015), they also make the opposite argument that primaries are less likely with “two relatively equal partners” (De Luca, Jones, and Tula 2002, 432). Similarly, Poiré (2002) argues that primaries are more likely when top contenders within a party are more evenly matched in terms of personal resources. As he puts it, when there is a clear frontrunner, party leaders can simply “use the old‐fashioned centralized process to select the obvious candidate” (Poiré 2002, 95). The mechanism driving this expectation is identical to the one illustrated in Figure 3; when the Maverick's quality (vM)$( {{v}_M} )$ falls outside the range (v̲,v¯)$(\underline{{v}} ,\bar{v})$, A's best course of action is straightforward—choose Appointment and then nominate the aspirant of higher quality.Before moving on, I emphasize an important distinction between this work and existing research.9Appendix Table A.1 (p. 13) summarizes the literature discussed in this section. Previous works have proposed either Hypothesis 1 or Hypothesis 2 but not both. To the best of my knowledge, this model is the first to advance the two hypotheses together and, most importantly, to provide a unified framework that accounts for these two empirical patterns. Moreover, because these hypotheses were generated from the same equilibrium (described in Proposition 1), I expect primaries to be predominantly used when both conditions are present, that is, when a party is electorally strong and the party's top two aspirants are of similar quality.Evidence from Mexican PrimariesI provide empirical support for the theory using data from Mexico. This is an ideal case to explore the implications of the model for two reasons. First, the assumption of weak politician–party linkages that motivates this research holds well. Although the levels of party switching and electoral volatility suggest party loyalties in Mexico are not as weak as they are in other countries,10For example, Brazil (Desposato 2006) or Peru (Levitsky 1999). party defections have been an important component of electoral politics in the country since at least the early 1990s (Langston 2002; Poiré 2002). Moreover, studies show party switching in Mexico is mainly driven by office‐seeking considerations and only exceptionally by ideological or policy‐related disagreements (Kerevel 2014). All this evidence is consistent with the existence of weak party–politician attachments.Second, major parties in the country have experimented with primaries at the federal and state levels for at least three decades (Bruhn 2010; Poiré 2002; Wuhs 2006). This is important because most empirical research on the determinants of nomination rules uses cross‐national data and exploits variation across parties or time (e.g., Lundell 2004). While these works can help identify how country‐level attributes shape the choice of these rules, they offer little leverage in terms of assessing the theoretical model. By contrast, the rich variation afforded by the Mexican case allows an examination of the party leaders' choices while accounting for any party‐ and election‐specific effects. Specifically, I can study how the same party leaders change their choices according to the conditions of each race, which is exactly the type of comparison that the theoretical model makes.The analysis consists of two parts. First, I analyze an original data set on candidate selection to the Chamber of Deputies (2003–2009). In each election during this period, two of the country's main parties, the PAN and the PRD, nominated a subset of their candidates in primaries, while the rest were appointed by their respective party leaders. I exploit this variation to study the factors that shaped the party leaders' decision to use primaries. Second, I study the determinants of gubernatorial primaries in the PRI during the last years of the party's hegemonic rule. For decades, gubernatorial candidates were handpicked by the president, but during the late 1990s the president decided to use primaries in several states (Díaz‐Cayeros and Langston 2004). I examine the choice of nomination rules across states during Ernesto Zedillo's presidency (1994–2000) using Poiré (2002)'s exceptional data set, which contains data from public opinion polls conducted by the Office of the Mexican Presidency at the time.Legislative Nominations, 2003–2009The Chamber of Deputies is elected every three years using a mixed‐member system. The chamber is formed by 500 legislators, of whom 300 are elected in single‐member districts by plurality rule and the remaining 200 by proportional representation. I focus on single‐member district nominations by the PAN and the PRD. In each election during 2003–2009, a subset of each party's nominees were selected in primaries, while the rest were appointed by the National Executive Committees (that is, the national leaders) of each party.1TableNomination Rules across Single‐Member DistrictsPartido de la Revolución Democrática (PRD)Partido Acción Nacional (PAN)200320062009Total200320062009TotalAppointment256215251722144140195479Primary448549178156160105421Total300300300900300300300900Notes: Distribution of nomination rules across single‐member districts in each legislative election during 2003–2009 by political party. Data from the IFE's General Council(see CG59/2003, CG76/2006, and CG173/2009).The other major party at the time, the PRI, is not included in the analysis due to the minimal variation in nomination rules within the same election: the party appointed six candidates in 2003, all 300 candidates in 2006, and seven candidates in 2009.11Moreover, even when the PRI purports to use primaries, in practice, candidates are selected by local elites. Wuhs (2006) argues that local leaders often stack the deck in favor of certain candidates, and Langston (2017) reports that governors control legislative nominations in their states. The lack of within‐election variation suggests that the choice of nomination rules was driven by election‐specific factors, offering little leverage to study the strategic choices of the PRI leaders in any given election. In contrast, the rich variation observed in the cases of the PAN and PRD makes it possible to study how electoral districts' characteristics shape the strategic decision making of the same party leaders within each party‐election.12The focus on the PAN and PRD is consistent with other work on legislative primaries in Mexico during this period (e.g., Bruhn 2010).In order to examine the factors that shaped the choice of primaries, I assemble a data set with information on the nomination rules used by each party in these elections. The unit of observation is the party–district–election. Below, I briefly describe the variables used in the analysis. Appendix B (pp. 14–16) provides details on data sources, variable operationalization, and descriptive statistics.The dependent variable is an indicator that takes the value of one if the party reported to the country's electoral commission, named Instituto Federal Electoral (IFE) at the time, that the candidate running in the district for the respective election was nominated in a primary and a value of zero otherwise. Table 1 shows the distribution of this variable by party and election year.According to Hypothesis 1, parties should be more likely to use primaries when their party is electorally stronger. The results presented below use the average of the party's district‐level vote share in the previous three legislative elections as the measure of strength. However, these results are very robust to the use of alternative measures, such as the party's district‐level vote share in the (1) previous legislative election, (2) most recent presidential election, and (3) previous mayoral election (see Appendix B, pp. 16–17). For reasons of data availability, I cannot test Hypothesis 2, which links the use of primaries to the relative quality of a party's top two aspirants. However, the analysis of gubernatorial candidate selection does test this hypothesis.The data set includes several covariates that, according to existing research, could explain the use of primaries. First, Mexican governors control key political resources that make them the de facto leaders of their parties' state branches, and can thus exert substantial influence over candidate selection within their parties (Rosas and Langston 2011) Therefore, I include an indicator of whether a district is in a state with a copartisan governor.Next, Poiré (2002) argues that concurrent local elections might decrease party leaders' incentives to use primaries since the availability of other offices allows them to compensate disgruntled aspirants. Thus, I include an indicator of whether a district is in a state where legislative and local elections were concurrent. Similarly, Bruhn (2010) argues that parties are less likely to hold primaries in rural districts where organizing a primary poses greater logistical challenges. I account for this using an index created by the Consejo Nacional de Población (National Population Council) that combines different dimensions of development using factor analysis. Finally, existing work suggests that the presence of incumbents could shape the decision to use primaries (Ichino and Nathan 2012). This consideration, however, does not apply to the Mexican case since, during the period of study, legislators were banned from running for consecutive reelection; that is, there were no incumbents.Table 2 reports the results of logit models with standard errors clustered by state. Model 1 uses pooled data and includes party–election fixed effects, whereas Models 2 and 3 are party‐specific and include election fixed effects. The results are extremely supportive of Hypothesis 1. Across all models, the coefficient of the measure of electoral strength is positive and statistically significant. The only other factor that systematically shapes the likelihood of a primary is having a copartisan governor (significant at 90% level), although this is only true for the PAN. To facilitate the interpretation, Figure 4 shows predicted probabilities of each party holding a primary as a function of their electoral strength in the district. As can be seen, both parties were significantly more likely to use primaries where they have obtained better electoral results.2TableDeterminants of Legislative Primaries in Mexico, 2003–2009PooledPartido Acción Nacional (PAN)Partido de la Revolución Democratica (PRD)Average vote share (previous three legislative elections)6.809**5.385**10.931**(0.880)(1.200)(2.668)Copartisan governor0.602†1.072†−0.135(0.350)(0.555)(0.759)Concurrent election−0.029−0.4270.466(0.185)(0.281)(0.355)Rural−0.194−0.104−0.509(0.141)(0.239)(0.325)Constant−2.324**−1.747**−5.516**(0.564)(0.360)(0.996)Party–election fixed effects√Election fixed effects√√N1,800900900Akaike Information Criterion1,764.1541,068.316655.717Notes: Cells report estimates of logit models with standard errors (clustered by state) in parentheses. The dependent variable is an indicator of whether party p held a legislative primary in district d at time t.†p < .1; *p < .05; **p < .01.4FigureElectoral Strength and Legislative PrimariesNotes: Predicted probabilities of primary by electoral strength (and 95% confidence intervals) estimated from Table 2, models (2)‐(3). All other variables held at their median values. Gray areas show histograms of the party's average vote share in the previous three elections. A potential drawback of the previous analysis has to do with the dependent variable, which only indicates whether a party reported to the IFE that it selected its candidate in a primary. However, Hypothesis 1 derives from Proposition 1, which gives necessary conditions for an equilibrium with a contested primary. Since the dependent variable potentially includes uncontested primaries, it is important to verify that the previous results hold, in particular, for contested primaries.To address this concern, I gather information on candidate entry in primaries. While the parties did not keep records of these races, the 2007 electoral reform created the legal figure of precandidato, which required any individual who aspired to get their party's nomination to register before the IFE (COFIPE 2008, Arts. 211–14). Critically, the registration process starts only after parties have informed the IFE about the nomination rules they will use. Thus, for the 2009 election, I use data on the registration of precandidatos to know whether a primary was contested.Table 3 replicates the previous models using an indicator of whether a party held a contested primary in the district as the dependent variable. Although this analysis is restricted to 2009, I am able to include two theoretically relevant controls that are not available for the entire 2003–2009 period. First, to account for the potential “stickiness” of primaries, I include an indicator of whether the party held a primary in the most recent legislative election (2006). Second, previous work suggests parties use primaries where their local branches are more powerful (Ichino and Nathan 2012). Thus, I include the logged number of party activists for every 1,000 registered voters in the district as a proxy for local party branch strength.3TableDeterminants of Contested Legislative Primaries in Mexico (2009)PooledPartido Acción Nacional (PAN)Partido de la Revolución Democratica (PRD)Average vote share (previous three legislative elections)4.106*4.655*12.686**(1.780)(2.296)(4.316)Copartisan governor1.255†2.090*−0.530(0.668)(0.911)(0.822)Primary in previous legislative election0.998*0.2761.788*(0.468)(0.486)(0.907)Concurrent election0.519−0.7761.695(0.508)(0.769)(1.113)Party activists (log)0.746**0.2750.613(0.247)(0.330)(0.419)Rural−1.234**−0.784†−2.001**(0.325)(0.421)(0.559)Constant−6.583**−4.740**−12.641**(0.860)(0.808)(1.833)Party fixed effects√N600300300Akaike Information Criterion360.036235.43494.035Notes: Cells report estimates of logit models with standard errors (clustered by state) in parentheses. The dependent variable is an indicator of whether party p held a contested legislative primary in district d in 2009.†p < .1; *p < .05; **p < .01.As before, the results are very supportive of Hypothesis 1. The coefficient of the measure of electoral strength is positive and statistically significant across all models. Even after controlling for the use of primaries in the previous election, parties are more likely to have contested primaries in districts where they are electorally dominant. This result is robust to the use of different measures of electoral strength (see Appendix B, pp. 18–19). Again, I find that the PAN is more likely to use primaries in states where they control the governorship. Finally, while there is no evidence that concurrent elections decrease the likelihood of primaries, these results do support Bruhn (2010)'s claim that parties avoid holding primaries in rural districts.Gubernatorial Nominations, 1994–2000Next, I study the determinants of gubernatorial candidate selection in the PRI during the presidency of Ernesto Zedillo (1994–2000). In the heyday of PRI dominance, nominations were controlled by the president, who as the de facto leader of the hegemonic party “had effective dedazo power at the state level, naming gubernatorial candidates and engaging in frequent governor‐switching” (Wuhs 2006, 44). However, starting in the 1990s, the new conditions of electoral competition created pressures for the party to consider alternative nomination methods. The first attempts were made under President Carlos Salinas (1988–1994), who in 1991 allowed the use of closed primaries in the states of Colima and Nuevo León (Poiré 2002). However, these processes were largely recognized as “sham exercises,” and in fact after those episodes, President Salinas “went back to centralized appointment for the rest of his term” (Poiré 2002, 102).It was not until President Zedillo's administration that the PRI consistently experimented with primaries to select its gubernatorial nominees (Díaz‐Cayeros and Langston 2004). Out of the 35 gubernatorial candidates nominated during Zedillo's tenure, 15 were selected via primaries (Poiré 2002). The empirical analysis that follows attempts to identify the factors that favored the use of primaries during this administration. It is important to stress that although the sample covers a six‐year period, the choice of nomination rules was always made by the same national leadership, that is, the presidency under Zedillo.13For a review of the PRI's formal nomination rules, see Poiré (2002, 106–8).I analyze Poiré (2002)'s unique data set, which combines information on the PRI's nomination rules with state‐wide polls conducted by the Office of the Mexican Presidency with the purpose of evaluating these gubernatorial contests. As Poiré explains, “each of these polls was taken some time before the nomination took place, and in every case, it served as an input to inform the presidency's decision on the choice of candidate and candidate selection procedure” (Poiré 2002, 118). Thus, although the sample is rather small, it provides a rare opportunity to analyze data used in the PRI's real‐life decision making.These rich polling data are used to measure the theory's key variables. First, I measure the PRI's electoral strength with a question that asked respondents about their vote intentions in the upcoming gubernatorial election. Because the polls were conducted before the nomination rules (and thus the candidates) were decided, this question should only reflect the popularity of the party in the state. Next, I measure the quality of the PRI's gubernatorial hopefuls with a question that asked, “Among these people, who would you prefer to be the PRI's candidate for governor in the coming election?” I use aggregate responses to this question to create an aspirant quality differential, calculated as the difference in the percentages received by the top two contenders. According to Hypothesis 2, primaries should occur when this differential is relatively low, indicating that the top two aspirants enjoy similar support among respondents.To start, I conduct difference‐in‐means tests. Consistent with Hypothesis 1, the PRI selected its gubernatorial candidates via primaries in states where the party's expected vote was higher. Among states where the PRI did and did not use primaries, the mean vote intention for the party was 44.52% and 39.35%, respectively. This difference is not statistically significant (p‐value = 0.12), which is expected given the small sample size. The evidence also supports Hypothesis 2. On average, the aspirant quality differential is more than two times larger in states where the PRI did not hold primaries than in states where it did (13.23 versus 6.34 percentage points). Notably, despite the sample size, this difference is on the edge of statistical significance (p‐value = 0.059).Additionally, Figure 5 presents a two‐by‐two classification of cases using dichotomous measures of PRI electoral strength and the aspirant quality differential.14For each variable, an observation is classified as “high” if its value is above the sample average and “low” otherwise. The focus is on the category in which both conditions established by the theory hold: support for the PRI is high (Hypothesis 1) and the aspirant quality differential is low (Hypothesis 2). Consistent with the theory, a majority of cases (50%) in which the PRI held primaries fall in this group, while the equivalent figure for the group in which the PRI did not hold primaries is 10.5%. Perhaps more informative is the comparison in the number of cases in each category across tables. When the two conditions of interest hold (two shaded cells), the PRI used primaries three times more often than not (six primaries versus two appointments). By contrast, the opposite pattern emerges when these two conditions do not hold. In each of the other categories, primaries were the exception rather than the rule (for more details see Appendix B, p. 19).5FigureDistribution of Cases by Nomination RuleNotes: Percentages calculated within each table. Number of cases appears in parentheses (n). To conclude, in Appendix B (pp. 22–28) I review existing accounts of the PRI's gubernatorial nominations under Zedillo and provide a short description of each primary during this period. This evidence is widely consistent with the mechanisms driving the adoption of primaries in the model. I summarize three key findings from this exercise. First, there is academic consensus that, in line with the model's main claim, the PRI leaders' motivation for using primaries was preventing potential defections (Poiré 2002; Reveles Vázquez 2000). As Díaz‐Cayeros and Langston explain[the challenge for] the PRI leadership was to choose gubernatorial candidates who could win clean elections, while not splitting the party at the state level … Under competitive elections, losers in the hegemonic party's nominations could leave the party, compete against it, and even defeat the “official” candidate, if an opposition party were willing to nominate them.(Díaz‐Cayeros and Langston 2004, 9–10)Second, the review of individual primaries shows that several aspirants (1) could have left the party during the race and secured another party's nomination and (2) probably would have left had the PRI's candidate been appointed. In fact, in many states where the PRI held primaries, it was an open secret that the opposition was trying to poach one of the party's gubernatorial hopefuls. For instance, according to a case study from the state of Sinaloa in 1998, the possibility of a defection not only explained the use of primaries but also:contributed to make the candidate selection process a fairly clean one … The PRD leaders had openly declared their willingness to nominate Millán [a contender in the PRI's primary] in case he decided to leave the PRI and join their party.(Reveles Vázquez 2000, 188–89)Finally, there is evidence that party leaders' control over the party's nominations triggered defections from aspirants who anticipated not receiving the party's ticket. One example of this dynamic is the case of Chiapas in 2000. In May 1999, then‐Senator Pablo Salazar left the PRI, arguing the governor would “try to impose his successor as governor” and announced he would work to “build a candidacy supported by regular citizens in order to run for the governorship in 2000” (Becerril 1999). Although the PRI ended up selecting the party's candidate in a primary (held in February 2000), by then Salazar had already secured the nomination of a broad pre‐electoral coalition that included eight opposition parties in the state (Mariscal 1999).ConclusionThis research is part of a growing literature that seeks to understand why parties democratize candidate selection. Most existing accounts consider settings in which electoral competition is programmatic, as is common in established democracies. In contrast, this article develops a new theory that focuses on the weak linkages between politicians and parties that characterize electoral politics in many developing countries and new democracies, where primaries are increasingly popular (Estaun 2015; Ichino and Nathan 2018).The theory centers around the party‐affiliation decision of an ambitious politician who, in the absence of meaningful partisan attachments, bases their choice on office‐seeking considerations alone. I show that by choosing a nomination rule, party leaders can affect the office‐seeker's career prospects and, thus, their incentives to stay in a political party. Nomination rules that centralize power in the hands of party leaders create a commitment problem that can trigger defections. Since the leaders cannot credibly promise to handpick some aspirants, a high‐quality contender who anticipates not being nominated will leave the party and attempt to run under a different ticket. Primaries emerge as a commitment device that, under certain conditions, allows party leaders to prevent costly defections.15This is consistent with Invernizzi's (2022) claim that by providing a more egalitarian distribution of electoral spoils, primaries can increase intraparty power‐sharing among factions.The main result is that parties strategically use primaries to prevent defections only when two conditions hold. First, the party is electorally strong ex ante. The reason for this is that an ambitious politician would only stay in a party that uses primaries—rather than joining one that hands them the nomination—if their chances of winning the general election upon winning the primary are high enough to compensate for the fact that they have to compete for the party's ticket. Second, the party's top two aspirants are of similar‐enough quality. This occurs for two reasons: (1) if the potential defector is high quality relative to the other aspirant, the commitment problem above disappears, and (2) if the potential defector is low quality compared to the other aspirant, the cost of a defection is low‐enough that the party leaders are willing to let this politician go.I provide empirical support for the theory using data from Mexico. Although the evidence comes from one country, there are several reasons why the theory and findings are more generally relevant for the study of candidate selection. Most importantly, the building blocks of the theory are not specific to Mexico. The model studies an election under two institutional features that are present in a wide range of democracies: (1) candidates compete in a single‐seat constituency (Bormann and Golder 2013), and (2) a party can choose its own nomination rules (Hazan and Rahat 2010). Additionally, the model assumes that the attachments between politicians and parties are weak. Thus, as mentioned, the theory should be more applicable to settings where this assumption holds (for example, where parties are not programmatic). Indeed, at least one of the model's expectations is consistent with findings from other settings in which these conditions hold well (Ichino and Nathan 2012; Kemahlioglu, Weitz‐Shapiro, and Hirano 2009).Finally, I also highlight that a particularity of the Mexican case—the ban on consecutive reelection—did not inform any modeling decisions. In fact, by having a one‐shot game, the model follows other formal accounts that study the choice of nomination rules in places where reelection is possible (e.g., Serra 2011; Snyder and Ting 2011). That said, extending the model to account for the repeated nature of elections could strengthen some of the strategic incentives driving the model's results. For instance, defections could become increasingly costly for party leaders. Specifically, when reelection is possible, failing to retain a high‐quality aspirant can have long‐term consequences. Similarly, for the Maverick, the cost of not receiving the party's ticket could increase. Even if running for office in the next election is possible, the need to compete against an incumbent and the emergence of new contenders could make winning office more challenging.ReferencesAdams, James, and Samuel Merrill. 2008. “Candidate and Party Strategies in Two‐Stage Elections Beginning with a Primary.” American Journal of Political Science 52(2): 344–59.Aldrich, John H. 1995. Why Parties? Chicago: The University of Chicago Press.Ascencio, Sergio J., and Yann P. Kerevel. 2021. “Party Strategy, Candidate Selection, and Legislative Behavior in Mexico.” Legislative Studies Quarterly 46(3): 713–743.Astudillo, Javier, and Klaus Detterbeck. 2020. “Why, Sometimes, Primaries? Intraparty Democratization as a Default Selection Mechanism in German and Spanish mainstream parties.” Party Politics 26(5): 594–604.Barrow, Lynda. 2007. “Politicians and Party Switching in Mexico.” Politics 27(3): 165–73.Becerril, Andrea. 1999. “Renuncia Salazar Mendiguchía al PRI ante miles de chiapanecos.” La Jornada. https://bit.ly/37XRb74.Bormann, Nils‐Christian, and Matt Golder. 2013. “Democratic Electoral Systems Around the World, 1946–2011.” Electoral Studies 32: 360–69.Bruhn, Kathleen. 2010. “Too Much Democracy? Primaries and Candidate Success in the 2006 Mexican Elections.” Latin American Politics and Society 52(4): 25–52.Bruhn, Kathleen. 2014. “Choosing How to Choose: From Democratic Primaries to Unholy Alliances in Mexico's Gubernatorial Elections.” Estudios Mexicanos 30: 212–40.Carey, John M., and John Polga‐Hecimovich. 2006. “Primary Elections and Candidate Strength in Latin America.” Journal of Politics 68(3): 530–43.Código Federal de Instituciones y Procedimientos Electorales (COFIPE). 2008. Diario Oficial de la Federación. January 14, 2008.Cox, Gary, and Matthew McCubbins. 1993. Legislative Leviathan: Party Government in the House. Cambridge: University of California Press.De Luca, Miguel, Mark P. Jones, and María Inés Tula. 2002. “Back Rooms or Ballot Boxes?: Candidate Nomination in Argentina.” Comparative Political Studies 35(4): 413–36.Desposato, Scott. 2006. “Parties for Rent? Ambition, Ideology, and Party Switching in Brazil's Chamber of Deputies.” American Journal of Political Science 50: 62–80.Desposato, Scott, and Ethan Scheiner. 2008. “Government Centralization and Party Affiliation: Legislator Strategies in Brazil and Japan.” American Political Science Review 102(4): 509–24.Díaz‐Cayeros, Alberto, and Joy Langston. 2004. “The Consequences of Competition: Gubernatorial Nominations and Candidate Quality in Mexico: 1994–2004.” https://bit.ly/2HMughyEstaun, Eva. 2015. “Primary Elections in Latin America.” In ACEProject Encyclopedia. https://aceproject.org/ace‐en/topics/pc/annex/pcy/primary‐elections‐in‐latin‐americaFashagba, Joseph. 2014. “Party Switching in the Senate under Nigeria's Fourth Republic.” The Journal of Legislative Studies 20(4): 516–41.Fell, Dafydd. 2013. “Impact of Candidate Selection Systems on Election Results: Evidence from Taiwan before and after the Change in Electoral Systems.” The China Quarterly March 213: 151–72.Hazan, Reuven Y., and Gideon Rahat. 2010. Democracy within Parties. Candidate Selection Methods and Their Political Consequences. Oxford: Oxford University Press.Heller, William, and Carol Mershon. 2005. “Party Switching in the Italian Chamber of Deputies, 1996–2001.” Journal of Politics 67(2): 536–59.Hicken, Allen. 2006. “Stuck in the Mud: Parties and Party Systems in Democratic Southeast Asia.” Taiwan Journal of Democracy 2(2): 23–46.Hix, Simon. 2004. “Electoral Institutions and Legislative Behavior: Explaining Voting Defection in the European Parliament.” World Politics 56: 194–223.Hortala‐Vallve, Rafael, and Hannes Mueller. 2015. “Primaries: The Unifying Force.” Public Choice 163(3): 289–305.Ichino, Nahomi, and Noah Nathan. 2018. “Primary Elections in New Democracies.” In Routledge Handbook of Primary Elections, ed. Robert G. Boatright. New York: Routledge.Ichino, Naomi, and Noah Nathan. 2012. “Primaries on Demand? Intra‐Party Politics and Nominations in Ghana.” British Journal of Political Science 42(4): 1–23.Invernizzi, Giovanna M. 2002. “Antagonistic Cooperation: Factional Competition in the Shadow of Elections.” American Journal of Political Science. https://doi.org/10.1111/ajps.12712Kemahlioglu, Ozge, Rebecca Weitz‐Shapiro, and Shigeo Hirano. 2009. “Why Primaries in Latin American Presidential Elections?” Journal of Politics 71(1): 339–52.Kerevel, Yann P. 2014. “Loyalty and Disloyalty in the Mexican Party System.” Latin American Politics and Society 56(3): 93–117.Langston, Joy. 2002. “Breaking Out Is Hard to Do: Exit, Voice, and Loyalty in Mexico's One‐Party Hegemonic Regime.” Latin American Politics and Society 44(3): 61–88.Langston, Joy. 2017. Democratization and Authoritarian Party Survival: Mexico's PRI. Oxford: Oxford University Press.Levitsky, Steven. 1999.“Fujimori and Post‐Party Politics in Peru.” Journal of Democracy 10(3): 78–92.Lundell, Krister. 2004. “Determinants of Candidate Selection: The Degree of Centralization in Comparative Perspective.” Party Politics 10: 25–47.Mainwaring, Scott and Timothy Scully, eds. 1995. Building Democratic Institutions: Parties and Party Systems in Latin America. Stanford CA: Stanford University Press.Mariscal, Ángeles. 1999. “Pablo Salazar, candidato PRD en Chiapas.” La Jornada. https://bit.ly/3kTXRWG.Poiré, Alejandro. 2002. Bounded Ambitions. Party Nominations, Discipline, and Defection: Mexico's PRI in Comparative Perspective. PhD diss., Harvard University.Reveles Vázquez, Francisco. 2000. “El PRI en las elecciones de Sinaloa.” Estudios Políticos Quinta época 23(23): 167–208.Rosas, Guillermo, and Joy Langston. 2011. “Gubernatorial Effects on the Voting Behavior of National Legislators.” Journal of Politics 73(2): 477–93.Serra, Gilles. 2011. “Why Primaries? The Party's Tradeoff between Policy and Valence.” Journal of Theoretical Politics 23(1): 21–51.Shabad, Goldia, and Kazimierz Slomczynski. 2004. “Inter‐party Mobility Among Parliamentary Candidates in Post‐Communist East Central Europe.” Party Politics 10(2): 151–76.Shomer, Yael. 2017. “The Conditional Effect of Electoral Systems and Intraparty Candidate Selection Processes on Parties Behavior.” Legislative Studies Quarterly 42(1): 63–96.Snyder, James, and Michael Ting. 2011. “Electoral Selection with Parties and Primaries.” American Journal of Political Science 55(4): 781–95.Wuhs, Steven. 2006. “Democratization and the Dynamics of Candidate Selection Rule Change in Mexico, 1991‐2003.” Estudios Mexicanos 22(1): 33–56.Zhirnov, Andrei, and Mariam Mufti. 2019. “The Electoral Constraints on Inter‐Party Mobility of Candidates: The Case of Pakistan.” Comparative Politics 51(4): 519–54.
Journal
American Journal of Political Science – Wiley
Published: Mar 13, 2023