This assignment is a take-home essay assignment of 2 questions, 2 pages each

Elections and Democratization in Authoritarian Regimes Author(s): Daniela Donno Source: American Journal of Political Science, Vol. 57, No. 3 (July 2013), pp. 703-716 Published by: Midwest Political Science Association Stable URL: https://www.jstor.org/stable/23496648 Accessed: 18-03-2020 15:40 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide

range of content in a trusted digital archive. We use information technology and tools to increase productivity and

facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

https://about.jstor.org/terms

Midwest Political Science Association is collaborating with JSTOR to digitize, preserve and extend access to American Journal of Political Science

This content downloaded from 66.77.17.54 on Wed, 18 Mar 2020 15:40:12 UTC All use subject to https://about.jstor.org/terms

Elections and Democratization in Authoritarian

Regimes

Daniela Donno University of Pittsburgh

When do elections in authoritarian regimes lead to democracy? Building from the distinction between competitive and hegemonic authoritarian regimes, I argue that presence of relatively weaker incumbents renders competitive authori

tarian elections more prone to democratization, but only when domestic and international actors choose to actively

pressure the regime. The effects of two forms of pressure—opposition electoral coalitions and international conditionality—

are theorized. Propositions are tested using a comprehensive dataset of elections in authoritarian regimes from 1990 to 2007. Results support two core claims: that the effect of electoral pressure is conditional on the type of authoritar

ianism and that this greater vulnerability to pressure is the reason why competitive authoritarian elections are more

likely to lead to democracy. In contrast, several alternative explanations—that differences across regime type are ex

plained by alternation in power, better electoral conduct, or ongoing processes of liberalization—are not supported by the evidence.

Since the end of the cold war, dictators around the globe have adapted to the changed international environment by adopting the form—though not

necessarily the substance—of democracy. The result has been a proliferation of electoral authoritarian regimes in

which political offices are filled through multiparty elec tions, but the electoral playing field is skewed in favor of the ruling party (Schedler 2006, 3). These "hybrid" regimes defy straightforward classification and chal lenge traditional, teleological theories of democratization (Diamond 2002). Their proliferation has therefore cre ated a surge of scholarly interest in explaining when they

democratize (Levitsky and Way 2010) and when dictators can be defeated at the ballot box (Bunce and Wolchik 2010a), as well as whether the repeated holding of elec tions produces democratization (Lindberg 2006, 2009). However, a central unresolved puzzle about electoral au thoritarian regimes is why elections serve to bolster au thoritarian rule at some times yet undermine it at others.

When do elections lead to democracy?

The first step toward answering this question is to consider the differences in electoral context across elec

toral authoritarian (EA) regimes: in hegemonic author itarian regimes (HARs), the incumbent or ruling party enjoys overwhelming electoral dominance (convention ally understood as winning more than 70 or 75% of the

vote or seat share); whereas in competitive authoritarian regimes (CARs), opposition parties pose greater electoral

challenges and garner a larger share of votes. Research has

shown that CARs are more likely to transition to democ

racy than HARs (Brownlee 2009; Roessler and Howard 2009). Yet, the underlying reasons for this difference— and the role played by elections—remain unclear. While it is clear that stronger opposition challengers in CARs should translate to a higher probability of alternation in power, it is not obvious why these regimes should be more likely to democratize—an outcome which entails a systematic improvement in the quality of elections. That many EA regimes remain durably authoritarian even af ter power changes hands attests to the importance of

Daniela Donno is Assistant Professor of Political Science, University of Pittsburgh, 4600 Posvar Hall, Pittsburgh, PA 15260 ([email protected]).

For helpful comments on earlier drafts, I am grateful to Despina Alexiadou, Thad Dunning, Steve Finkel, Michael Goodhart, Julia Gray, Ken Greene, Marc Morjé Howard, Susan Hyde, Judith Kelley, Staffan Lindberg, Scott Morgenstern, Anibal Pérez-Linan, Nita Rudra, Burcu Savun, Susan Stokes, Leonard Wantchekon, as well as participants at the Duke University Seminar on Global Governance and Democracy; the Juan March Institute Workshop on Electoral Fraud, Vote Buying and Clientelism; and the University of Pittsburgh Globalization Workshop. Additional thanks to Jason Brownlee, Marc Howard, Philip Roessler, Judith Kelley, Susan Hyde, and Nikolay Marinov for generously sharing their data. Shawna Metzger, Daniel Tirone, and Bruno Hoepfers provided excellent research assistance.

American Journal of Political Science, Vol. 57, No. 3, July 2013, Pp. 703-716

©2013, Midwest Political Science Association DOI: 10.1111/ajps. 12013

703

This content downloaded from 66.77.17.54 on Wed, 18 Mar 2020 15:40:12 UTC All use subject to https://about.jstor.org/terms

704 DANIELA DONNO

treating turnover and democratization as distinct phe nomena. Moreover, the presence of stronger opposition challengers in CARs only increases the incumbent's incen tives to manipulate elections—an outcome antithetical to democratization.

While existing research has tended to study elec toral politics in competitive and hegemonic authoritarian regimes separately, this article forwards a unified frame work for understanding elections across EA regimes. It takes the key difference between CARs and HARs—the degree of incumbent or ruling party dominance—as the starting point for theorizing the conditions under which

authoritarian elections lead to democratization. I argue that two key forms of electoral pressure are more effective in CARs than in HARs. First, the relative weakness of in

cumbents in competitive authoritarian elections affords

opposition parties with greater opportunities to forge electorally viable coalitions, which, in turn, decrease the

incumbent's ability to engage in electoral manipulation. Coalitions in HARs, in contrast, are unlikely to spur im proved electoral conduct and democratization because they are not perceived as a threat. Second, incumbents' weaker electoral standing in CARs means that their need for external support is greater, increasing their sensitiv ity to international pressure for democracy. By compari son, hegemonic regimes are more insulated from external pressure. Thus, although opposition parties coalesce and international actors impose conditionality at nearly iden tical rates in hegemonic and competitive authoritarian elections, these forms of pressure are only effective in CARs.

To test these propositions, I analyze the full set of elec

tions in EA regimes (both competitive and hegemonic) from 1990 to 2007. Information on opposition coalitions and international conditionality is originally coded. By focusing on the application of specific tools of interna tional pressure for democracy, this study also differs from

previous research on structural, diffuse forms of interna tional influence, such as economic linkage or membership in international organizations.

The results of the analysis strongly support the claim

that opposition coalitions and international conditional ity are effective levers for democratization in CARs, but not in HARs. I then evaluate a number of alternative ex

planations for why competitive authoritarian elections are

more likely to lead to democracy. I find, perhaps surpris

ingly, that CARs are no more likely than HARs to be on a

liberalizing path to democracy, nor do they exhibit better

electoral conduct. Moreover, CARs' greater propensity to

democratize is not explained by their higher frequency of alternation in power. These facts, which belie con ventional wisdom about electoral authoritarian regimes,

imply that competitive authoritarian elections are not inherently destabilizing. Rather, it is the application of specific forms of domestic or international pressure that can transform these contests into mechanisms for de

mocratization. When applied in HARs, the same forms of pressure are ineffective.

Electoral Authoritarian Regimes

The distinction between electoral authoritarianism and

democracy hinges on the quality of electoral competition (Diamond 2002; Schedler 2006, chap. 1 ). EA regimes allow

multiple parties to compete in elections, but they do so under patently unfair conditions. Incumbents may place

barriers on opposition parties' ability to campaign; gener ate a progovernment media bias; stack electoral commis

sions and courts with their supporters; or resort to stuffing

ballot boxes and manipulating vote tabulations. Among EA regimes, a further distinction can be made based on

the degree to which the incumbent or ruling party is elec

torally dominant (Brownlee 2009, 518; Diamond 2002). In HARs, the ruling party wins elections by wide margins,

while in CARs, elections serve as "arenas through which opposition forces may—and frequently do—pose signif icant challenges" (Levitsky and Way 2002, 54), despite many obstacles to their success. Election results matter be

cause they signal the regime's level of control over the po

litical arena (Brownlee 2007, 3), as well as (more noisily) its level of popular support (Gandhi 2008, 167; Gandhi and Lust-Okar 2009, 405). Overwhelming electoral vic tories therefore contribute to an aura of invincibility that

discourages elite defections and deters voters from sup porting the opposition (Simpser 2013). Thus, the use of incumbent vote shares to distinguish between HARs and CARs is not merely a convenient analytical shortcut; elec tion results are a direct measure of authoritarian political dominance.

To understand why some authoritarian regimes are hegemonic and others more competitive, one must look to a set of deep-rooted institutional and economic con ditions.1 Some regimes that secure political dominance are able to do so because of their resource advantages— stemming, for example, from a large, politicized pub lic sector, or natural resource wealth—that are used to

maintain patronage networks and buy mass support (Magaloni 2006, 15-24). These resource advantages, in turn, create an incentive structure that induces the

1 This is different from the question of regime survival. An autocracy may prove highly durable despite an inability to secure hegemony.

This content downloaded from 66.77.17.54 on Wed, 18 Mar 2020 15:40:12 UTC All use subject to https://about.jstor.org/terms

ELECTIONS AND DEMOCRATIZATION 705

opposition to pursue self-defeating electoral tactics (Greene 2007). Other accounts focus on ruling parties as mechanisms for maintaining broad support coalitions (Brownlee 2007; Magaloni 2008), on the early repression of labor movements as a source of enduring opposition weakness (LeBas 2011, Chap. 2), or on the role of class based communal conflict, which can spur elites to sup port the creation of an authoritarian "Leviathan" (Slater

2010). In short, a regime's degree of electoral hegemony— which determines its status as a HAR or CAR—is shaped

by path-dependent factors with deep institutional roots.

The structure of the economy, colonial legacies, and elite choices made during the early years of the regime cru cially shape its ability to consolidate power in the longer term.

Transitions to Democracy

For electoral authoritarian regimes—which hold multi party elections but rig the playing field—democratization entails an improvement in the quality of elections such that parties compete on an equal footing, the casting and

counting of ballots is conducted "in the absence of mas sive voter fraud," and results "are representative of the public will" (Freedom House 2010). It is important to underscore that democratization involves a change in the quality and conduct—not necessarily the outcome—of elections. Many EA regimes experience changes in lead ership, even an opposition victory, yet exhibit no subse quent improvement in the quality of electoral competi tion (Levitsky and Way 2010, 21).

Existing accounts of democratization in EA regimes leave several questions unanswered. In their study of competitive authoritarianism in the post—cold war era, Levitsky and Way (2010) argue that economic and social linkage with the West (which varies by region) is the pri mary cause of democratization. Yet, linkage is a diffuse, structural form of influence that fails to capture the ef fects of concrete tools of international pressure targeted to certain countries at certain times. Indeed, one of the

limitations of Levitsky and Way's theory is that there is no

pathway through which active pressure—which they term

international leverage—can lead to democracy (Slater 2011). Moreover, their claims about long-term regime trajectories are unable to explain the particular timing of democratic transitions—which is, in practice, almost

always linked to elections. Quantitative evidence clearly suggests the impor

tance of electoral politics for democratization. Roessler and Howard (2009) and Brownlee (2009) both find that

the probability of a democratic transition is closely related

to the regime's prior electoral dominance. But this finding

raises further questions about causal mechanisms. While it is straightforward to understand why electorally weaker incumbents in CARs should be more likely to lose power

(cf. Roessler and Howard 2009, 103), we lack a satisfy ing explanation for why CARs should be more likely to democratize—which, to reiterate, entails a systematic im provement in the quality of elections. Brownlee suggests

that the reasons stem from the "higher levels of con testation enjoyed by opposition parties" in CARs (2009, 521), yet he notes the need for "closer examination of the

causal processes that propel these trends" (530). How ex actly does the existence of greater electoral contestation in CARs translate to democratization? And how do electoral

dynamics differ in HARs?

In a pathbreaking analysis of 50 competitive author itarian elections, Howard and Roessler (2006) find that

liberalizing electoral outcomes are most likely when op position parties mobilize and forge coalitions.2 But re search on hegemonic regimes suggests a fundamentally different electoral dynamic. Rather than providing op portunities for genuine political competition, elections in HARs serve primarily as mechanisms to determine ac cess to patronage (Lust-Okar 2009), to manage intra-elite

competition (Blaydes 2011), or to signal the incumbent's strength (Gandhi and Lust-Okar 2009, 407). Under these conditions, democratization is caused not by proximate electoral strategies but by changes in the underlying dis tribution of resources—for example, in response to an economic downturn or a shrinking state presence in the economy—which, over time, reduces the regime's ability

to maintain popular support and elite unity (Reuter and Gandhi 2011). Elections themselves rarely pose a threat,

and when they do, they are likely reflecting these deeper structural changes.

In what follows, I build from these insights, explain

ing how two forms of pressure—opposition coalitions and international conditionality—help produce democ ratization in CARs but are unlikely to be effective in HARs.

Opposition Coalitions

To orchestrate electoral misconduct, incumbents re

quire cooperation. Party operatives, election commission

members, polling station workers, police, and the me dia are those most commonly complicit—either actively

2Bunce and Wolchik (2010a) similarly emphasize the importance of opposition strategies in ousting cheating incumbents (a different dependent variable) in postcommunist CARs.

This content downloaded from 66.77.17.54 on Wed, 18 Mar 2020 15:40:12 UTC All use subject to https://about.jstor.org/terms

J06 PANIFIA DONNO

or passively—in electoral manipulation. The choices of these actors depend crucially on perceptions about who is likely to win the election. When opposition parties are

fragmented and weak—as in most EA regimes—there is little doubt about electoral outcomes, and orders to or

chestrate or tolerate misconduct are likely to be followed with little hesitation. Domestic actors will simply calculate

that their best option is to side with the regime (Bratton and Van de Walle 1997; Van de Walle 2006). If opposition parties unite, however, this can lead to a dramatic transformation in perceptions. For those who must choose whether to engage in, or tolerate, mis

conduct, the decision is no longer so clear-cut. As Hale notes in his analysis of post-Soviet presidential elections, if the incumbent is believed to be on his way out, regimes

will "face defection from their very own machines," in

cluding the "elites controlling mass media and even the very courts that are necessary for consummating electoral

fraud" (2005,141). Howard and Roessler further explain that in the presence of an opposition coalition, ".. .the police, army, and bureaucrats may be less inclined to em ploy illegal practices to benefit the incumbent" (2006, 371).

An important caveat, however, is that opposition coalitions will only have such an effect when they are electorally viable, meaning they are perceived as having a

real chance of victory. Opposition parties in CARs begin from a position of relative strength compared to those in hegemonic contexts; they have a better track record of electoral performance, which often translates to greater

representation in the legislature. For these reasons, coali tions in CARs are far more likely to gain enough traction to be perceived as electorally viable. In contrast, opposi tion coalitions in HARs are typically composed of weak parties with poor organizational cohesion and low levels of popular support.

To illustrate this point, consider the different ef fects of coalition building in Serbia and Belarus. Dur ing the 1990s, Serbia was an archetypical CAR. Freedom was highly curtailed and elections unfair, but Slobodan Milosevic was never able to consolidate electoral dom

inance. Although the ruling party's vote share ranged from just 29-53% during the 1990s,3 a fragmented op position consistently failed to mount a credible elec toral challenge.4 The situation changed dramatically in the run-up to the 2000 presidential election when op

3These figures increase if one accounts for other parties allied with the government, but the total progovernment vote share never exceeded two-thirds of the vote.

4See Cohen (2002) on the opposition's repeated electoral failures in the 1990s.

position factions set aside their differences and formed a broad coalition—the Democratic Opposition of Ser bia (DOS)—with Vojislav Kostunica as their presidential candidate. The coalition had a profound impact on public

perceptions, spurring unprecedented pre-election mobi lization, campaign activity, and voter turnout (Bunce and Wolchik 2010b). While Milosevic did everything in his power to manipulate the contest in his favor, ultimately,

defections among key actors in the state (including the army, interior ministry, and police), media, and even his own party prevented him from consummating outcome changing fraud.5

Similar coalition-building efforts by the opposition in Belarus fell flat. Different from Serbia—where oppo sition forces were always politically relevant despite their

disadvantage—President Aleksander Lukashenka in Be larus enjoys near absolute power. Since his ascent in 1994, elections have been little more than a farce. Presidential

contests in 2001 and 2006 were won by overwhelming margins, and by 2004, there was no opposition presence in the legislature. Under these conditions, the Belarusian

opposition's attempt to mimic the Serbian revolution by uniting behind a single presidential candidate in 2001 (and again in 2006) failed to gain any traction (Boris 2001; Shargorodsky 2001; Silitski 2010). Without a tip ping point in perceptions about the opposition's chances

of victory, the regime's grip over electoral management bodies, the media, and security forces remained firm, and

the majority of citizens remained too afraid, or simply too passive, to defend their right to vote in free and fair elections.

HI: The effect of opposition coalitions on democ ratization is greater in competitive authoritar ian elections than in hegemonic authoritarian elections.

International Conditionality

Since the end of the cold war, the international com

munity has become increasingly active in democracy promotion. There are many tools in the democracy promotion "toolkit"—including election monitoring, democracy aid, and diplomatic pressure—but the most direct and immediate way to exert pressure for free and fair elections is through conditionality, understood as the linking of concrete punishments or rewards to improvements in the quality of elections. "Negative"

5See Pavlakovic (2005, 128), Graham (2000), Glenny (2000), and Sandford (2000).

This content downloaded from 66.77.17.54 on Wed, 18 Mar 2020 15:40:12 UTC All use subject to https://about.jstor.org/terms

ELECTIONS AND DEMOCRATIZATION 707

conditionality threatens either material (e.g., economic sanctions) or political (e.g., suspension of mem bership in international organizations) costs on the government for electoral misconduct, while "positive" conditionality holds out the promise of rewards if con duct improves. International conditionality can also have indirect consequences for a regime, since demo cratic credentials are now an important factor influ encing the allocation of foreign aid, international in vestment, and multilateral loans (Hyde 2011; Lebovic and Voeten 2009). Moreover, conditionality can impose symbolic costs, particularly if a government bases its claims to legitimacy on its international support and its electoral credentials. By activating these material and symbolic incentives, conditionality can induce govern ments to reduce their reliance on electoral misconduct

or to tie their hands by introducing institutional re forms that make misconduct more difficult and risky (Donno 2013).

Different regimes exhibit different levels of sensitivity

to international pressure, however. For incumbents in an

already tenuous electoral position, a loss of international support can have serious consequences, particularly in the short term. Withdrawal of economic benefits in the

months leading up to an election is likely to be particularly

costly for CARs, for this is the time when resources are most needed to help them win, buy, or steal votes. In con trast, incumbents in HARs are better insulated from inter

national pressure. Patronage networks sustained through

domestic sources—natural resource wealth, a large pub lic sector, or state-controlled economy (as in Belarus, or Mexico under the PRI)—are unaffected by international pressure for democracy. Even when external actors do wield leverage over a hegemonic regime, the leadership can typically find ways to introduce cosmetic reforms that do not threaten its power base.6 Equally important, the symbolic consequences of international criticism are likely to be less severe for HARs. Where elections are bla tantly lopsided contests that serve primarily to determine

access to state resources, citizens do not expect them to be free and fair or evaluate them on this basis (Lust Okar 2009). If there is no illusion of democracy to begin with, international efforts to delegitimize the regime on

these grounds are unlikely to be effective. Many hege monic leaders can simply dismiss international criticism

altogether, as did Lukashenka in Belarus when he urged supporters in 2001 to "take the election monitors by the

scruff of the neck and send them packing so they can fly

off with their heads spinning" (Boris 2001).

6See, for example, Blaydes' (2011, chap. 10) account of Mubarak's response to U.S. pressure for multicandidate presidential elections.

H2: The effect of international pressure for democracy is greater in competitive authoritarian elections than in hegemonic authoritarian elections.

In sum, pressure for clean elections occurs across regime types. Indeed, as the next section shows, oppo sition parties coalesce and international actors impose conditionality at nearly identical rates in hegemonic and competitive authoritarian regimes. But it should be more effective in CARs. This is so not because competitive au thoritarian leaders are closet democrats who willingly for

swear electoral fraud. Rather, the application of pressure

in CARs alters incentives and beliefs, triggers defections, and prompts institutional reforms that limit the govern ment's ability to manipulate elections.

H3: Elections in CARs are more likely to lead to democracy than elections in HARs because CARs are more vulnerable to domestic and international

pressure.

Data

To uncover the factors associated with democratization

through elections, I constructed a dataset of elections in EA regimes from 1990 to 2007. The units of analysis are national elections, both presidential and legislative. Creation of the dataset proceeded in several steps. First, Geddes, Wright, and Franz's (2012; GWF hereafter) data were used to identify the set of autocratic regime-years. These data, which cover all countries in the world from

1946 to 2010, expand and improve upon Geddes' (1999) earlier coding.7 Within this set, I define the sample of electoral authoritarian regimes as those in which multi ple parties are allowed to contest elections. If the ruling party or candidate won less than 75% of the votes (presi dential contests, first round) or seats (legislative elections)

in the last election, the regime is classified as a CAR; oth erwise, it is classified as a HAR.8 This coding reflects the country's status as of January 1 of the year in question,

7Data and codebook are available at <http: //dictators.la.psu .edu/ >. The GWF data are highly correlated with Cheibub, Gandhi, and Vreeland's (2010) coding of regime type, but GWF has the ad vantage, for my purposes, of explicitly considering electoral quality in its coding of autocracy.

8Following Brownlee (2009), using the World Bank's Database of Political Institutions (DPI) indexes of executive and legislative elec toral competitiveness (Keefer 2010), a country is coded as CAR if it received a "7" on either index and as HAR if its highest score on either index was a "5" or a "6." These scores capture the pres ence of multiple parties in an election and whether the winner ex ceeded 75% of the votes (presidential elections) or seats (legislative

This content downloaded from 66.77.17.54 on Wed, 18 Mar 2020 15:40:12 UTC All use subject to https://about.jstor.org/terms

γθ8 DANIELA DONNO

thus equivalent to a one-year lag.9 Last, I employed the dataset on National Elections across Democracy and Au tocracy (NELDA) to identify the elections held in these regimes (Hyde and Marinov 2012). This yields a sample of 177 elections (104 in CARs and 73 in HARs). A list of

included elections and regime type is in the supplemental appendix.

The dependent variable of the analysis is a transition

to electoral democracy. I employ a dummy variable indi cating whether a country moved from a "0" to a "1" on Freedom House's list of electoral democracies in the year of the election in question. Inclusion on this list requires

that a country meet four criteria:

1) a competitive, multiparty political system;

2) universal adult suffrage for all citizens (with ex ceptions for restrictions that states may legiti mately place on citizens as sanctions for criminal offenses);

3) regularly contested elections conducted in condi tions of ballot secrecy, reasonable ballot security, and in the absence of massive voter fraud, and that

yield results that are representative of the public will; and

4) significant public access of major political parties to the electorate through the media and through generally open political campaigning.

The advantages of this measure are twofold. First, un

like the Polity index, which incorporates nonelectoral di mensions in its coding, inclusion on the electoral democ

racy list is driven by electoral quality alone (with criteria 3

and 4 representing the main differences between an elec toral authoritarian regime and a democracy). Second, unlike other dichotomous measures of regime type, in clusion on the list is independent of electoral outcomes.10 Of the 177 elections in the data, 25 marked a transition

elections). Remaining regimes (scoring a 4 or lower on both in dexes) are coded as closed authoritarian.

9If a country is coded as closed authoritarian as of January 1, that country-year is excluded from the dataset. This coding scheme therefore excludes founding elections in which a country made a rapid transition from a closed to a multiparty system and then held an election in the same year. Appendix F in the supplementary materials presents the results of robustness checks that alter this coding scheme by ( 1 ) including all founding multiparty elections following single-party rule and then (2) excluding all founding multiparty elections in which the regime is coded as a HAR or CAR as of January 1. The results hold in both cases.

10Cheibub, Gandhi, and Vreeland (2010) and Geddes, Wright, and Franz (2012) both require that a country experience alternation in power in order to move from dictatorship to democracy.

to electoral democracy; 18 of these occurred in CARs and seven in HARs.11

The first key independent variable is an indicator for opposition coalition. Following Howard and Roessler (2006), this measure is coded as "1" if all major opposi tion parties forged a unified platform, coordinated their campaigns, or united behind a single presidential candi date.12

The second key independent variable, which cap tures the application of international pressure, is an indi cator for pre-election conditionality, defined as the is suance of threats or promises that link punishments or rewards to the country's electoral conduct (Donno 2013).13 Incentives can be economic (e.g., sanctions, aid, trade agreements) or political (e.g., suspension of diplo matic ties, suspension or granting of membership in an international organization). While previous research on international influences on democracy has focused largely on passive forces, such as international linkage (Levitsky and Way 2010) or membership in democratic regional organizations (Pevehouse 2005), these data rep resent a systematic collection of information on ac tive pressure for democracy. The conditionality variable is coded as "1" if one or more of the following ac tors employed conditionality during the four months prior to the election: the United States, United Nations,

European Union (EU), Organization for Security and Cooperation in Europe (OSCE), Council of Europe, Organization of American States (OAS), Caribbean Com munity (CARICOM), Southern African Development Community (SADC), Economic Community of West African States (ECOWAS), and African Union.14

An initial look at the data reveals that the rates

of coalitions and conditionality across regime types are nearly equal: coalitions occurred in 15% of elections in both HARs and CARs, while international conditionality

"See Appendix Β in the supplementary materials for a list.

12Howard and Roessler's data were used when available. See Appendix C in the supplementary materials for details on how the remaining elections were coded.

13Conditionality is only coded if clearly targeted toward the country and election in question; regional or ongoing conditionality policies not related to that election are not coded.

14These actors, which were identified through extensive research of a larger set of international organizations, represent the universe of regional organizations with a track record of at least once employing conditionality for election-related purposes. The primary sources of information used were stories in international newspapers and newswires. To ensure reliability, all elections were coded twice by different individuals working independently. Intercoder disagree ment was found in less than 10% of observations and was resolved

by the author. See Appendix C in the supplementary materials for more information.

This content downloaded from 66.77.17.54 on Wed, 18 Mar 2020 15:40:12 UTC All use subject to https://about.jstor.org/terms

ELECTIONS AND DEMOCRATIZATION 709

was applied in 8% of hegemonic elections and 13% of competitive authoritarian elections (this difference was not statistically significant). There appears to be little ground for concern, then, that opposition parties and international actors are systematically choosing to focus their efforts in the ostensibly easier, more competitive, cases.

Seven control variables capture country- and election-specific factors that may influence the proba bility of a democratic transition. First, it is likely that democratization occurs primarily in elections that deter

mine who will hold executive power. Because the stakes are higher in these contests, they are more likely to elicit

higher levels of domestic mobilization for democracy. I include a dichotomous variable, main election, coded as

a "1" for presidential elections in presidential (or mixed)

systems and legislative elections in parliamentary systems

(Simpser and Donno 2012). Second, I include a vari able indicating whether the incumbent was running in the

election, which is expected to decrease the chances of de mocratization (Cheeseman 2010; Maltz 2007).15 It is also

important to account for the country's previous experi ence with elections. Lindberg (2006, 2009) argues that holding elections fosters institutional change and greater

respect for civil liberties and that these changes cumulate over time. Relatedly, Bratton and Van de Walle (1997) ar

gue that the more elections held under authoritarianism, the greater the chances for democratization. I thus in clude a variable that sums the number of previous elections

held under a continuous authoritarian spell.16 Two variables control for economic conditions: first

a variable for GDP per capita, lagged one year.17 High income is reliably associated with democracy, though its effect on democratic transitions is less clear. Przeworski

et al. (2000, chap. 2), for example, find that dictatorships at high (but not the highest) levels of income are more likely to democratize. Second, I include a variable for GDP growth, measured as the percent change in a coun try's GDP from year t-2 to t-1. If it is true that good eco

nomic performance bolsters authoritarian regimes, the coefficient on this variable should be negative.

15Data come from the NELDA dataset (Hyde and Marinov 2012), which defines the incumbent leader according to the individual identified in Archigos (Goemans, Gleditsch, and Chiozza 2009).

16This count begins in 1946 or at the date of independence. The count restarts each time a country transitions to autocracy from democracy. Data on authoritarian spells are from Geddes, Wright, and Franz (2012) and data on elections are from NELDA (Hyde and Marinov 2012).

17 Data on GDP and GDP growth are taken from the World Bank (2010).

Finally, Levitsky and Way (2010) demonstrate that democratization is more likely in regions of the world with

high economic and social linkage with the West—namely, Latin America and Central/Eastern Europe. Accordingly, I include dummy variables for both these regions in the models.

Findings

To recap, two core implications flow from the argument. First, opposition coalitions and international condition ally should be more effective at producing democrati zation through elections in CARs than in HARs. Thus, interaction terms between competitive authoritarianism and these two forms of pressure are expected to be positive

and statistically significant. Second, this greater vulner ability to pressure explains why CARs are more likely to democratize than HARs. Thus, any significant differ

ence in the probability of democratization across regime types should disappear once the conditional effects of op position coalitions and international conditionality are modeled. To put these predictions to the test, I estimate probit models with democratic transition as the depen dent variable. The sample includes all elections in EA regimes, excepting only those countries already classified as electoral democracies by Freedom House,18 as well as countries under a provisional government installed after

the resignation of an authoritarian leader. All models are run with robust clustered standard errors to adjust for

dependence across observations within countries.19 The first column of Table 1 presents a baseline model

that sheds first light on the conditions under which au thoritarian elections lead to democracy. The results reveal that democratization is, as expected, more likely in main elections that determine who holds executive power, but less likely when the incumbent is running for reelection, lending support to the idea that open-seat elections rep resent particularly opportune moments for democratic change (cf. Cheeseman 2010). The results also suggest that strong economic performance bolsters authoritar ian regimes, since high income and economic growth are

18These ambiguous cases in which GWF and Freedom House dis agree on status as an autocracy include, for example, elections in Botswana (all years), Haiti (2000), and Russia (1995-2004). I omit these cases from the main analysis, but when they are included, the results for competitive authoritarianism, opposition coalitions, and international conditionality remain robust.

"Because the data form an unbalanced panel, panel-corrected stan dard errors are infeasible. In a likelihood ratio test, country random effects were not significantly different from zero, indicating that a pooled model is preferred. Fixed effects lead to too many observa tions dropping from the analysis.

This content downloaded from 66.77.17.54 on Wed, 18 Mar 2020 15:40:12 UTC All use subject to https://about.jstor.org/terms

JIO DANIELA DONNO

Table 1 The Determinants of Democratization through Elections

(1) (2) (3) (4) (5)

Competitive Authoritarian Regime (CAR) 0.64+ 0.58+ 0.18 0.70* 0.21

(0.05) (0.08) (0.62) (0.03) (0.57)

Opposition Coalition 0.69+ -4.21** -4.12**

(0.09) (0.00) (0.00)

International Conditionality 0.18 -4.88** -4.62**

(0.73) (0.00) (0.00)

CAR x Opp. Coalition 5.15** 5.45**

(0.00) (0.00)

CAR x Conditionality 5.23** 5.34**

(0.00) (0.00) Main Election 1.60** 1.59** 1.65** 1.61** 1.77**

(0.00) (0.00) (0.00) (0.00) (0.00)

Incumbent Running -0.79* -0.80* -0.88* -0.84* -0.87*

(0.02) (0.02) (0.02) (0.02) (0.03) # Previous Elections -0.01 -0.01 -0.01 -0.05 -0.07+

(0.82) (0.77) (0.74) (0.22) (0.09) GDP per capita (lagged 1 year) -0.42* -0.46* -0.42+ -0.37+ -0.37+

(0.04) (0.03) (0.06) (0.05) (0.07) GDP growth (lagged 1 year) -0.04+ -0.04+ -0.04 -0.03 -0.02

(0.07) (0.09) (0.12) (0.14) (0.54) Central and Eastern Europe 0.32 0.18 0.13 0.56 0.44

(0.49) (0.72) (0.80) (0.36) (0.47) Americas 1.45** 1.63** 1.58** 1.40** 1.63**

(0.00) (0.00) (0.00) (0.01) (0.00) Alternation, Previous Elec. 0.46 1.05+

(0.38) (0.09) Misconduct Intensity -0.34+ -0.54*

(0.06) (0.02) Regime Openness 0.03 0.06

(0.90) (0.78) Prior Liberalization 0.31* 0.37+

(0.03) (0.05) Constant 0.28 0.42 0.49 0.49 0.58

(0.83) (0.77) (0.75) (0.73) (0.72) Pseudo R2 0.28 0.31 0.35 0.34 0.43 Observations 177 177 177 176 176

Note: Robust p-values in parentheses. +significant at 10%; *significant at 5%; "significant at 1%.

negatively associated with democratization. However, the

idea that the repeated holding of elections helps produce democracy is not supported.20

20Lindberg (2006) shows that the holding of elections is associ ated with democratic change in Africa, but this relationship does not appear to hold in Latin America (McCoy and Hartlyn 2009) or the Middle East (Lust-Okar 2009); moreover, the effect of re peated elections is strongest when held in more democratic contexts (Teorell and Hadenius 2009, 96-98).

In line with previous research, Model 1 shows that de

mocratization is more likely in competitive authoritarian

elections than in hegemonic elections. Even after control

ling for a host of country- and election-specific factors,

there is something different about elections in CARs— some attribute associated with democratization—that re

mains unexplained. Models 2 and 3 resolve this puzzle. The variables for opposition coalition and international conditionality are positively signed (Model 2) but are only statistically significant when interacted with the indicator

This content downloaded from 66.77.17.54 on Wed, 18 Mar 2020 15:40:12 UTC All use subject to https://about.jstor.org/terms

ELECTIONS AND DEMOCRATIZATION γη

Table 2 Predicted Probability of Democratization

Probability of Democratic Transition

Competitive Authoritarian Regimes No Coalition, No Conditionality .26

[.09, .50] Coalition .59

[.21, .92]

Conditionality .39

[.05, .84]

Coalition + Conditionality .69

[.22, .97]

Hegemonic Authoritarian Regimes No Coalition, No Conditionality .21

[.07, .43] Coalition .00

[.00, .00]

Conditionality .00

[.00, .00]

Coalition + Conditionality .00

[.00, .00]

Note: The 95% confidence intervals appear in brackets. Other vari ables held at modal (for dummy variables) or median values.

for competitive authoritarianism. Thus, opposition coali tions and international conditionality greatly increase the likelihood of democratization, but only in CARs, where governments are more vulnerable to electoral pressure.21 This resonates with Wright's (2009) finding that foreign

aid conditionality is more effective in regimes with larger

winning coalitions (measured in part by the competitive ness of executive selection). Notably, once the interaction terms are included in the models, the constituent term

for CARs becomes statistically insignificant, lending sup port to the claim that vulnerability to pressure is the reason why elections in CARs are more likely to lead to democracy (Hypothesis 3). Absent an opposition coali tion or international conditionality, democratization in a competitive authoritarian context is no more likely than

democratization in a hegemonic context.

To illustrate the core findings, Table 2 presents the predicted probability of democratization through elec tions in CARs and HARs, all other factors held equal.22 In

21 Results are robust when the conditionality variable is recoded to include U.S. and EU conditionality only.

22Table 2 is based on simulations performed using Clarify (Tomz, Wittenberg, and King 2001). Unless otherwise specified, all vari ables are held at their median values.

an average hegemonic authoritarian election with no op

position coalition and no international conditionality, the probability of democratization is 0.21. This probability is only slightly higher—at 0.26—in an identical competi tive authoritarian election. The fates of CARs and HARs

diverge, however, with the application of domestic and international pressure. When opposition parties coalesce or international actors impose pre-election condition ality in CARs, the chances of democratization increase to 60 and 40%, respectively; and in elections with both coalitions and conditionality, the probability of democ ratization climbs to 0.69.

The results also reveal that in hegemonic regimes, coalitions and conditionality are negatively associated with democratization (denoted by the coefficients on the constituent terms in Model 3). This is consistent with the argument forwarded here, that these forms of pres

sure should not have a positive effect in HARs. Yet it suggests the intriguing possibility that they are actually

counterproductive when applied in hegemonic contexts, perhaps because leaders respond by further tightening their grip on power. In Cambodia's hegemonic regime, for example, sustained international involvement has ar

guably only spurred Hun Sen to further centralize power and manipulate electoral rules so that outright vote rig ging is unnecessary.23 The negative findings for coalitions

and conditionality in hegemonic regimes should be in terpreted with caution, however, since there are relatively few cases of democratization in HARs in the data. Fu

ture qualitative research will need to probe whether these effects hold in particular cases over time.

Alternative Explanations

This article's core hypothesis, that CARs are more vulner able to opposition and international electoral pressure than HARs, is not the only possible explanation for why competitive authoritarian elections are more likely to lead

to democracy. Elections in CARs and HARs may differ in other ways that are causally relevant for democratization.

First, as previous research has noted, CARs are more likely

than HARs or closed regimes to experience transfers of

executive power to the opposition (Roessler and Howard 2009). Because alternation in power is an electoral out come that temporally follows (rather than precedes) any

improvement in electoral conduct, it cannot be said to exert a causal effect on democratization in the current

23Brinkley (2011) gives an in-depth account of the international community's role in Cambodia.

This content downloaded from 66.77.17.54 on Wed, 18 Mar 2020 15:40:12 UTC All use subject to https://about.jstor.org/terms

γΐ2 DANIELA DONNO

election; but it may increase the chances of democrati zation in the next election if the new government ex hibits greater respect for democratic norms. To explore this possibility, I employ a dummy variable for alternation

in executive power in the previous election. In presidential

systems, alternation is coded if the pro-government can7 didate (including, if applicable, the incumbent himself) lost a presidential election; in parliamentary systems, if the ruling party lost a legislative election.24

A second set of alternative explanations stems from the idea that CARs may be closer to democracy to begin with. Elections in CARs, though flawed, may be marked

by relatively better conduct than HARs. If so, the changes

required to cross the threshold to democratic elections would be smaller and, possibly, easier to achieve. To eval uate this, I construct a measure for the intensity of electoral

misconduct. Ranging from 1 to 3, this variable sums the number of areas in which misconduct is present. For each

election, one point is given for pre-election restrictions on the opposition's freedom of movement, association, or expression; one point for pre-election bias in the media or

institutions that govern elections; and one point for ballot

fraud. These forms of misconduct are identified using a

variety of sources, including more than 400 election ob servation reports, news articles and the NELDA dataset.25 Additionally, elections in CARs may simply represent the culmination of an ongoing process of liberalization. While competitive authoritarianism is a stable equilibrium for some countries, for others, it may represent a transitional

point on the road from a closed or hegemonic regime to democracy. Following Howard and Roessler (2006), the extant level of regime openness is captured using the country's Freedom House civil liberties score (measured as the running average in the two years prior to the elec tion).26 Prior liberalization is measured as the difference

in the Freedom House political rights score during the four-year period preceding the election. An initial look at the data provides only limited sup

port for these arguments, calling into question some con ventional assumptions about differences across hybrid regimes. Alternation in power is indeed more likely in CARs: 18% of executive elections in CARs led to alter

nation, compared to 13% in HARs. And alternation is associated with a higher chance of democratization in the

24This variable is taken from NELDA (question 24; Hyde and Mari nov 2012). Missing data were filled in using Goemens, Gleditsch, and Chiozza's (2009) coding of regular alternation in power.

25 See Appendix C in the supplementary materials for a more de tailed description of this variable.

26Scores are inverted so that higher values represent greater free dom.

Figure 1 Intensity of Electoral Misconduct, by Authoritarian Type

next election (33% versus 10% in contests not following alternation), but this difference is not statistically signif

icant.27 Elections in CARs are more likely to occur in a more open context, with an average civil liberties score in the two years prior to an election of 3.28 in CARs, com pared to 2.94 in HARs.28 However, elections in CARs are not more likely to occur during periods of ongoing politi cal liberalization. The average change in Freedom House's

political rights index in the four years prior to elections is

negative in both regime types (—0.06 in HARs and —0.20 in CARs), and this difference is not significant.

Turning to the quality of elections, Figure 1 shows the percent of elections in each regime type marked by different levels of misconduct intensity. It reveals, strik

ingly, that elections in CARs are actually marked by more widespread misconduct—in terms of the scope and

range of tools employed—than elections in HARs. This lends support to the idea that CARs rely to a greater extent on outright electoral manipulation, while HARs are typically able to sustain themselves without resorting to these tactics (Lust-Okar 2009; Magaloni 2006). Elec tions in HARs are, however, more likely to exhibit prob

lems in their legal framework, such as barriers on who can run for office, high thresholds for party registration,

and flaws in procedures for lodging electoral complaints (Kelley 2009, 2010).29 In sum, elections in CARs are, by

27 A chi2 test yields a p-value of .44; the null hypothesis of indepen dence cannot be rejected.

28A difference of means test was significant at ρ = .00.

29In Kelley's (2010) quality of elections (QED) data, 66% of elec tions in HARs are marked by moderate or major problems in their structural/legal environment, compared to 51% in CARs.

This content downloaded from 66.77.17.54 on Wed, 18 Mar 2020 15:40:12 UTC All use subject to https://about.jstor.org/terms

ELECTIONS AND DEMOCRATIZATION 713

definition, more competitive in terms of their outcome

and tend to be governed by relatively better legal frame works, but they are not more free and fair in terms of their conduct.

Models 4 and 5 (Table 1) explore the effect of these

variables in the multivariate analysis. The results show that elections marked by more intense misconduct are, as

expected, less likely to democratize, while prior liberaliza

tion is positively associated with democratization. Alter nation in power exhibits the expected positive effect only

in Model 5, and it is only marginally significant. Most importantly, the coefficient on CARs remains at nearly

identical levels of statistical significance after the inclu sion of these variables (Model 4 versus Model 1), which means that the four alternative explanations account for a

very small amount of the variance across CARs and HARs.

The interactive effects of coalitions and conditionality are also unaffected.

Robustness

The robustness of this article's core findings is probed in several ways. I first investigate whether the results hold

under an entirely different coding scheme for regime type and democratization. Roessler and Howard (2009) code EA regimes as those that hold multiparty elections and that receive a Freedom House political rights score of 3 or

higher and a Polity score of 5 or lower. (Democratization

occurs if the country moves to an FH score of 2 or bet ter or a Polity score of 6 or higher.) Within EA regimes, they employ an electoral threshold of 70% (rather than 7 5%) of votes/seats to the winner for classifying a country

as a CAR or HAR. When this alternative coding scheme is employed, the key results remain nearly unchanged (supplementary Appendix D), the only difference being that the effect of opposition coalitions in CARs reduces to marginal levels of statistical significance. Upon closer examination, this appears to be driven by a subset of cases where executive elections are won by large margins, but legislative elections are more competitive (coded as hegemonic by Roessler/Howard but coded as competi tive in the main analyses). This suggests that opposition

coalitions are particularly effective in settings where the executive is electorally dominant but the opposition gains

a greater foothold in the legislature.

To further probe the sensitivity of the results to dif

ferent electoral thresholds, I run a set of analyses that dispense with the dichotomous classification of CARs and HARs altogether and instead employ a continu ous measure of vote/seat shares, which I label authori

tarian electoral dominance.30 This variable captures the vote/seat share of the winning candidate/party in the last election. While this measure contains some missing data, and measurement error is more problematic than with the dichotomous classification of CARs and HARs,

the results again prove robust (Table Ε and Figure Ε in the supplementary materials). As dominance increases, the odds of democratization decline. Most notably, coalitions and conditionality exhibit significant in teractive effects: at low to mid-levels of electoral

dominance—analogous to competitive authoritarian regimes—coalitions and conditionality are statistically significant, but their positive effect on democratization

disappears in systems with highly skewed electoral results.

As a last check for sensitivity to different electoral thresholds, I ensured that the findings are not driven by differences in electoral system. The vote and seat shares of winning parties are typically smaller in proportional representation (PR) systems (in which larger numbers of parties compete and gain representation) than in majori tarian systems. However, inclusion of an indicator for PR

does not alter the results: competitive authoritarianism retains a positive, significant coefficient in the baseline model, and the conditional effects of coalitions and con

ditionality remain. Finally, I explored the possibility of selection bias.

As discussed, a large body of research has identified the

sources of authoritarian hegemony as being rooted in do mestic institutional legacies, economic endowments, and

the structure of the economy. Nevertheless, to investigate

statistically whether status as a CAR is driven by an un derlying, unobserved propensity for democratization, I employ a Heckman selection model that predicts both (a) selection into competitive authoritarianism and (b) democratization. The method, model specification, and results are discussed in the supplementary appendix. In short, the findings lend no support to the idea that se lection bias is driving the results, further supporting the

conclusion that opposition coalitions and international conditionality indeed exert a causal effect in CARs.

Conclusion

Democratization through elections has become an in creasingly prominent mode of regime change (Lindberg

2009). Yet, puzzlingly, the study of democratization in hybrid regimes has remained largely disconnected from

30This measure is taken from electoral results reported in Keefer (2010).

This content downloaded from 66.77.17.54 on Wed, 18 Mar 2020 15:40:12 UTC All use subject to https://about.jstor.org/terms

γΐ4 DANIELA DONNO

the growing literature on electoral conduct and quality. The few studies that have focused squarely on electoral politics have been limited in scope, analyzing a subset of elections in competitive authoritarian regimes (cf. Bunce and Wolchik 2010a; Howard and Roessler 2006). This

study has helped bridge this divide by examining a com prehensive sample of elections in EA regimes since the end of the cold war. In the process, it has challenged some commonly held ideas about hybrid regimes. Con tra those who tout the intrinsic tensions and fragility of competitive authoritarianism, the findings here un derscore that there is nothing inherently unstable about this type of hybrid regime. Instead, CARs are more ac curately characterized as potentially unstable in that de mocratization is contingent on whether domestic and international actors choose to pressure the regime. Ab sent pressure, elections in CARs are no more likely to lead to democracy than elections in hegemonic regimes.

Nor are elections in CARs manifestly closer to democracy to begin with, at least in terms of their conduct. While hegemonic authoritarian elections tend to be held under an inferior legal framework, active electoral misconduct

is more widespread in CARs, likely because leaders in these regimes have a greater need for manipulation to ensure victory. Yet, in CARs, the government's ability to manipulate can be offset if opposition parties forge a uni fied front and international actors threaten to punish the

regime for violations of electoral norms. Thus, in contrast

to the conclusion that international leverage produces, at best, unstable authoritarianism (Levitsky and Way 2010), this article has demonstrated that there is a link between

conditionality and democratization. Further, the find ing that the effect of international pressure is conditional

on regime context also helps explain why external ac tors have previously been noted as having only a "vari able impact" on electoral politics (Bunce and Wolchik 2010a).

This study also challenges analysts to think carefully about the relationship between democratization—which is related to the electoral process—and alternation in power—which relates to electoral outcomes. While these

two phenomena often coincide—for example, the democ

ratizing elections in Indonesia ( 1999), Croatia (2000), and

Peru (2001)—they should not be conflated. Indeed, the finding that alternation is not a significant predictor of

democratization in the next election suggests that alter nation may best be understood as a by-product, rather than a direct cause, of democracy. This point has im plications for domestic and international groups inter ested in democracy promotion, for it underscores that a focus on ousting dictators should be accompanied in equal measure by close attention to the democratic pro

cess, particularly the quality of elections. The electoral revolution in Kyrgyzstan is a stark example of alter nation in power unaccompanied by any real improve ment in democratic performance, but the revolutions in

Georgia (2004) and Ukraine (2005) exhibit a similar dy namic. Eight years on, both countries continue to strug gle with electoral irregularities, declining media freedom,

and, in Ukraine, a troubling upswing in opposition in timidation. In contrast, alternation in power in Slovakia

(1998) was accompanied by rapid institutional reform, facilitated by conditionality and assistance from the Eu ropean Union.

Variation across these postcommunist cases may prove instructive for the recent revolutions in the Arab

world, where euphoria over regime change must now give way to more practical consideration of the chal lenges associated with ensuring free and fair elections and,

in the longer term, achieving wholesale political trans formation. Elections will surely play an integral role in

these countries' transitions, but the findings presented here underscore the importance of the political context

in which these contests are held. In long-standing hege monic regimes, where citizens have lived through decades

of lopsided electoral charades, the creation of a compet itive political system with a robust opposition will be es

sential for altering public perceptions about the purpose, authenticity, and legitimacy of elections.

References

Blaydes, Lisa. 2011. Elections and Distributive Politics in Mubarak's Egypt. New York: Cambridge University Press.

Boris, Joseph. 2001. "If Belarus Vote Is Free and Fair, So What?" United Press International. September 9.

Bratton, Michael, and Nicolas Van de Walle. 1997. Democratic Experiments in Africa: Regime Transitions in Comparative Perspective. Cambridge: Cambridge University Press.

Brinkley, Joel. 2011. Cambodia's Curse: The Modern History of a Troubled Land. New York: Public Affairs.

Brownlee, Jason. 2007. Authoritarianism in an Age of Democra tization. New York: Cambridge University Press.

Brownlee, Jason. 2009. "Portents of Pluralism: How Hybrid Regimes Affect Democratic Transition." American Journal of Political Science 53(3): 515-32.

Bunce, Valerie J., and Sharon L. Wolchik. 2010a. "Defeating Dictators: Electoral Change and Stability in Competitive Au thoritarian Regimes." World Politics 62(1): 43-86.

Bunce, Valerie J., and Sharon L. Wolchik. 2010b. "Defining and Domesticating the Electoral Model: A Comparison of Slo vakia and Serbia." In Democracy and Authoritarianism in the Postcommunist World, ed. V. Bunce, M. McFaul, and K. Stoner-Weiss. New York: Cambridge University Press, 134-54.

This content downloaded from 66.77.17.54 on Wed, 18 Mar 2020 15:40:12 UTC All use subject to https://about.jstor.org/terms

ELECTIONS AND DEMOCRATIZATION 715

Cheeseman, Nie. 2010. "African Elections as Vehicles for Change." Journal of Democracy 21(4): 139-153.

Cheibub, José Antonio, Jennifer Gandhi, and James Raymond Vreeland. 2010. "Democracy and Dictatorship Revisited." Public Choice 143: 67-101.

Cohen, Lenard J. 2002. Serpent in the Bosom: The Rise and Fall of Slobodan Milosevic. Boulder, CO: Westview Press.

Diamond, Larry. 2002. "Thinking about Hybrid Regimes." Jour nal of Democracy 13(2): 21-35.

Donno, Daniela. 2013. Defending Democratic Norms: Interna tional Actors and the Politics of Electoral Misconduct. New York: Oxford University Press.

Freedom House, various years. "Freedom in the World: Political Rights and Civil Liberties." http: //www.freedomhouse.org (accessed November 2010).

Gandhi, Jennifer. 2008. Political Institutions under Dictatorship. New York: Cambridge University Press.

Gandhi, Jennifer, and Ellen Lust-Okar. 2009. "Elections under Authoritarianism." Annual Review of Political Science 12: 403-22.

Geddes, Barbara. 1999. "What Do We Know about Democrati zation after Twenty Years?" Annual Review of Political Science 2: 115-44.

Geddes, Barbara, Joseph Wright, and Erica Frantz. 2012. Authoritarian Regimes: A New Data Set. Unpublished manuscript, Penn State University.

Glenny, Misha. 2000. "Milosevic Allies Admit Defeat." The Times (London), September 30.

Goemans, Hein, Kristian Gleditsch, and Giacomo Chiozza. 2009. "Introducing Archigos: A Data Set of Political Lead ers." Journal of Peace Research 46(2): 269-83.

Graham, Bob. 2000. "Army and Church Tell Milosevic He's Beaten." Daily Mail, September 29.

Greene, Kenneth. 2007. Why Dominant Parties Lose: Mexico's Democratization in Comparative Perspective. New York: Cambridge University Press.

Hale, Henry E. 2005. "Regime Cycles: Democracy, Autoc racy, and Revolution in Post-Soviet Eurasia." World Politics 58(October): 133-65.

Howard, Marc Morjé, and Philip Roessler. 2006. "Liberalizing Electoral Outcomes in Competitive Authoritarian Regimes." American Journal of Political Science 50(2): 365-81.

Hyde, Susan D. 2011. The Pseudo-Democrat's Dilemma. Ithaca, NY: Cornell University Press.

Hyde, Susan D., and Nikolay Marinov. 2012. "Which Elections Can Be Lost?" Political Analysis 20(2): 191-210.

Keefer, Philip. 2010. Database of Political Institutions. Washing ton, DC: The World Bank.

Kelley, Judith. 2009. "D-Minus Elections: The Politics and Norms of International Election Observation." International

Organization 63(4): 765-87.

Kelley, Judith. 2010. "Quality of Elections Data (QED)." Duke University, http://www.duke.edu/web/diem/data.html.

LeBas, Adrienne. 2011. From Protest to Parties: Party-Building and Democratization in Africa. Oxford, UK: Oxford Univer sity Press.

Lebovic, James Η., and Erik Voeten. 2009. "The Cost of Shame: International Organizations and Foreign Aid in the Punish ing of Human Rights Violators." Journal of Peace Research 46(1): 79-97.

Levitsky, Steven, and Lucan A. Way. 2002. "The Rise of Compet itive Authoritarianism." Journal of Democracy 13(2): 51-65.

Levitsky, Steven, and Lucan A. Way. 2010. Competitive Author itarianism: Hybrid Regimes after the Cold War. New York: Cambridge University Press.

Lindberg, Staffan I. 2006. Democracy and Elections in Africa. Baltimore: Johns Hopkins University Press.

Lindberg, Staffan I., ed. 2009. Democratization by Elections: A New Mode of Transition. Baltimore: Johns Hopkins Univer sity Press.

Lust-Okar, Ellen. 2009. "Legislative Elections in Hegemonic Authoritarian Regimes: Competitive Clientelism and Resis tance to Democratization." In Democratization by Elections: A New Mode of Transition, ed. S. Lindberg. Baltimore: Johns Hopkins University Press, 226-45.

Maltz, Gideon. 2007. "The Case for Presidential Term Limits." Journal of Democracy 18(1): 128-42.

Magaloni, Beatriz. 2006. Voting for Autocracy: Hegemonic Party Survival and Its Demise in Mexico. New York: Cambridge University Press.

Magaloni, Beatriz. 2008. "Credible Power-Sharing and the Longevity of Authoritarian Rule." Comparative Political Studies 41(4): 715-41.

McCoy, Jennifer, and Jonathan Hartlyn. 2009. "The Relative Powerlessness of Elections." In Democratization by Elections: A New Mode of Transition, ed. by S. Lindberg. Baltimore: The Johns Hopkins University Press, 47-76.

Pavlakovic, Vjeran. 2005. "Serbia Transformed? Political Dy namics in the Milosevic Era and After." In Serbia since 1989:

Politics and Society under Milosevic and After, ed. S. P. Ramet and V. Pavlakovic. Seattle: University of Washington Press, 13-54.

Pevehouse, Jon C. 2005. Democracy from Above? Regional Orga nizations and Democratization. New York: Cambridge Uni versity Press.

Przeworski, Adam, Michael E. Alvarez, Jose Antonio Cheibub, and Fernando Limongi. 2000. Democracy and Development: Political Institutions and Well-Being in the World, 1950-1990. Cambridge: Cambridge University Press.

Reuter, Ora John, and Jennifer Gandhi. 2011. "Economic Per formance and Elite Defection from Hegemonic Parties." British Journal of Political Science 41(1): 83-110.

Roessler, Philip G., and Marc M. Howard. 2009. "Post-Cold War Political Regimes: When Do Elections Matter?" In De mocratization by Elections, ed. Staffan Lindberg. Baltimore: Johns Hopkins University Press, 101-27.

Sandford, Gillian. 2000. "Reporters Rebel against Milosevic: Media Call for Truthful Coverage Leads to Sacking." The Guardian (London), October 4.

Schedler, Andreas, ed. 2006. Electoral Authoritarianism: The Dy namics ofUnfree Competition. Boulder, CO: Lynne Rienner.

Shargorodsky, Sergei. 2001. "Foes Call Belarus Leader Dictator." Associated Press, September 8.

This content downloaded from 66.77.17.54 on Wed, 18 Mar 2020 15:40:12 UTC All use subject to https://about.jstor.org/terms

yi6 DANIELA DONNO

Silitski, Vitali. 2010. "Contagion Deterred: Preemptive Authori tarianism in the Former Soviet Union (the Case of Belarus)." In Democracy and Authoritarianism in the Postcommunist World, ed. V. Bunce, M. McFaul, and K. Stoner-Weiss. New York: Cambridge University Press, 274-299.

Simpser, Alberto. 2013. Why Governments and Parties Manip ulate Elections: Theory, Practice and Implications. New York: Cambridge University Press.

Simpser, Alberto and Daniela Donno. 2012. "Can International Election Monitoring Harm Governance?" Journal of Politics 74(2): 501-13.

Slater, Dan. 2010. Ordering Power: Contentious Politics and Au thoritarian Leviathans in Southeast Asia. New York: Cam

bridge University Press.

Slater, Dan. 2011. "Review: Competitive Authoritarianism: Hy brid Regimes after the Cold War." Perspectives on Politics 9(2): 385-88.

Teorell, Jan, and Axel Hadenius. 2009. "Elections as Levers of Democratization." In Democratization by Elections: A New Mode of Transition, ed. S. Lindberg. Baltimore: Johns Hop kins University Press, 77-100.

Tomz, Michael, Jason Wittenberg, and Gary King. 2001. CLARIFY: Software for Interpreting and Presenting Statistical Results. Version 2.0. Cambridge, MA: Harvard University.

Van de Walle, Nicolas. 2006. "Tipping Games: When Do Oppo sition Parties Coalesce?" In Electoral Authoritarianism: The

Dynamics of Unfree Competititon, ed. A. Schedler. Boulder, CO: Lynne Rienner, pp. 74-94.

World Bank. 2010. "World Development Indicators." www.worldbank.org/data.

Wright, Joseph. 2009. "How Foreign Aid Can Foster Democ ratization in Authoritarian Regimes." American Journal of Political Science 53(3): 552-71.

Supporting Information

Additional Supporting Information may be found in the online version of this article at the publisher's web site:

Supplementary Appendix A. Elections in HARs and CARs, 1990-2007

Supplementary Appendix B. Transitions to Electoral

Democracy in HARs and CARs, 1990-2007 Supplementary Appendix C. Coding Rules and

Sources

Supplementary Table D. Table 1 Models with Roessler and Howard (2009) Regime Type

Supplementary Table E. Predicting Democratization with a Continuous Measure of Authoritarian Electoral

Dominance

Supplementary Figure E. Effect of Coalitions and Conditionality on Democratization, Conditional on Au thoritarian Electoral Dominance

Supplementary Appendix F. Robustness Checks on Different Samples

Supplementary Table Fl. Replication of Results on Less Restrictive Sample

Supplementary Table F2. Replication of Results on More Restrictive Sample

Supplementary Appendix G. Selection Model of De mocratization in CARs 52

Supplementary Table G. Heckman Selection Model of Democratization in CARs

This content downloaded from 66.77.17.54 on Wed, 18 Mar 2020 15:40:12 UTC All use subject to https://about.jstor.org/terms