Too Many to Fight: Elite Free Riding and the Origins of Personalism

Abstract

Why do powerful elites in authoritarian regimes so often fail to resist autocratic power grabs? Existing scholarship suggests that strong elites can deter ambitious dictators and thereby prevent personalization. This paper qualifies that view. It argues that a strong but numerous ruling coalition can facilitate power grabs. When many actors have the capacity to constrain the leader, each has an incentive to free ride on the resistance of others. While all regime members benefit from a constrained executive, the costs of resistance are borne individually. As a result, regime members are often tempted to stand by and let others bear those costs. Thus, although power sharing is on average a constraining force on dictators, it can also demobilize elites and open the door to personalization. We must therefore ask not only how much power dictators share, but also how they share it.

Keywords

personalism autocracy free riding collective action coups

Authoritarian leaders are seldom satisfied with sharing power. They often seek to monopolize it to secure their rule and escape elite constraints. Such power grabs typically succeed when regime elites are too weak to resist them. Yet personalization is not limited to these cases. The current wave of personalism has affected the majority of dictatorships, including those where once-powerful elites should have been able to resist it (Chin et al., 2021; Kendall-Taylor et al., 2017). Why do regime elites with the capacity to constrain dictators so often fail to prevent the consolidation of power?

This question remains surprisingly underexplored. Existing scholarship has made significant strides in explaining why personalization occurs, but most accounts emphasize elite weakness as its primary cause. Fragile or fragmented ruling coalitions, scholars argue, cannot effectively constrain dictators and instead fall prey to purges or institutional marginalization (Bokobza et al., 2022; Escribà-Folch et al., 2020; Geddes et al., 2018; Sudduth, 2017). Similarly, theories of endogenous institutionalization suggest that elites who have secured a prominent position in the regime should be able to retain their influence, since confronting them is too costly for the leader (Magaloni, 2008; Meng, 2020; Reuter, 2017; Svolik, 2012). This suggests a stable equilibrium: weak elites stay weak, while strong ones remain strong. What they do not explain, however, is how powerful elites get marginalized.

And yet, personalization does occur in countries with strong ruling coalitions. In Indonesia, Suharto gradually eroded the autonomy of the military, transforming it from a co-equal pillar of the regime into a subordinate instrument of presidential control (Slater, 2009). The Mexican PRI, once the archetype of a hegemonic party, was gradually marginalized by presidents who learned how to undermine intra-elite alliances (Kaire, 2025). In Russia, Vladimir Putin initially faced a powerful constellation of elites—oligarchs, security-service figures, and Yeltsin-era technocrats—but he steadily neutralized them, recentralizing power around himself (Baturo and Elkink, 2021). These cases and others show that personalization can occur even under powerful ruling coalitions.

Some work has started to address the question of personalization under strong elites. This research often focuses on exogenous shocks, such as windfalls from natural resources or foreign aid, that provide autocrats with the resources they need to circumvent elite bargaining (Boutton, 2019; Casey, 2020; Fails, 2020). Others focus on the strategies dictators employ to gain dominance (Slater, 2003; Timoneda, 2020; Woldense, 2022). While these accounts offer important insights, they focus on the actions and opportunities of dictators. They tell us less about why elites themselves fail to coordinate in defense of their position.

In contrast, this article focuses on personalization as a problem of elite coordination and free riding. I argue that as the number of powerful actors within the ruling coalition increases, the incentives for individual members to resist power grabs decline. When the ruling coalition is numerous, individual members can plausibly expect that, if they do not check the dictator, someone else will (Olson, 1965). This logic of free riding can open the door to personalization, defined here as the degree to which the dictator centralizes decision-making authority (Baturo, 2024; Escribà-Folch and Timoneda, 2024; Slater, 2009).

Autocratic Personalization as Elite Free Riding

Even in the cutthroat world of authoritarian politics, it is in the collective interest of regime elites to act together to prevent the concentration of power. While regime members may occasionally benefit from the downfall of their peers—seizing promotions, rents, or a larger proportion of power—everyone in the ruling coalition is safer when the dictator remains constrained. A leader who must worry about elite pushback is less likely to act against their ruling coalition. As autocrats consolidate power, they shed these constraints and become increasingly insulated from their inner circle. This leaves all regime members more vulnerable. Even those who once served as staunch allies of the dictator may eventually find themselves isolated and the next target of the leader’s ambitions.

But resisting the dictator is rarely costless. Even when elites recognize the long-term danger of personalization, confronting the autocrat can come with high and immediate risks. Challenging the leader may invite political retaliation, loss of access to rents, and physical danger. Resistance is a risky bet: if it fails, the challengers may be jailed, exiled, or worse. As a result, even when all regime members stand to benefit from constraining the ruler, any one of them may prefer to let others take the lead, hoping to enjoy the benefits of collective action without paying its costs.

This free-riding problem lies at the heart of personalization under strong elites, understood here as regime insiders with the capacity to oust the autocrat (Boix and Svolik, 2013; Kaire, 2024; Myerson, 2008). The benefits of elite resistance resemble a club good—non-rivalrous and non-excludable among regime insiders. Once the dictator is constrained, all elites enjoy the resulting stability and security, regardless of whether they contributed to the effort. But the costs of resisting fall on individual actors. Each elite therefore faces a rational temptation to remain on the sidelines and let others bear the costs of opposition. Doing so might be too risky when only a handful of elites can constrain the leader. In the extreme, it would not make sense for the only strong elite in the regime to free ride on the resistance of others. But the temptation to do so increases when there are many others in the regime who could in principle resist the leader. Hence, a strong but numerous ruling coalition may be more susceptible to the leader’s power grabs than a relatively weaker but smaller elite.

This argument then joins others in rethinking the consequences of the way dictators choose to share power. Earlier work had argued that dictators who share power narrowly would be better off because it would be easier to buy the loyalty of their inner circle through targeted rents (Bueno de Mesquita et al., 2003, similarly Bausch, 2017). Subsequent research challenged that view, suggesting that larger ruling coalitions have a harder time coordinating because they are more susceptible to factionalism (Arriola, 2009; Geddes et al., 2018; Kaire, 2024; Marcum and Brown, 2016). This article builds on these latter studies and shows that large ruling coalitions can enable personalization but not only because of factionalism. Even if elites are unified in their policy and political preferences, the incentive to free ride will remain if the coalition includes a sufficient number of powerful members. Factionalism is not necessary for autocracies with strong but large ruling coalitions to devolve into personalist regimes.

Hypotheses

I have argued that the power of one autocratic elite discourages others from resisting the leader. However, empirically evaluating this claim presents two challenges. First, autocratic regimes differ widely in institutional structure and elite composition. To ensure comparability, I focus on two of the most common and politically significant elite types: party elites and military elites (Lai and Slater, 2006). While not present in every dictatorship, these actors appear frequently enough across regimes to permit meaningful cross-national analyses.

The second challenge is identifying observable indicators of elite resistance. Elite dissatisfaction can manifest in subtle and often opaque ways—such as foot-dragging, information hoarding, or undermining clientelist networks. These forms of resistance are difficult to detect and distinguish from simple ineffectiveness. To overcome this, I follow recent research and use reshuffling coup attempts as an empirical proxy for elite resistance (Chin et al., 2021; Goemans et al., 2009; Powell and Thyne, 2011). These coups, led by regime insiders with the intent of replacing the leader while preserving the regime, provide a direct and observable signal of dissent within the ruling coalition.

Given the above, I first evaluate the argument through two complementary hypotheses. The key prediction is that when one elite group is strong, the other is more likely to defer, expecting that resistance will be handled by others. Consequently:

H1a: The probability of a military reshuffling coup attempt decreases when party elites are strong.

H1b: The probability of a party-led reshuffling coup attempt decreases when military elites are strong.

In addition to these two main hypotheses, I also explore a secondary one. If personalization advances due to elite inaction, higher costs of resistance should also open the door to power grabs. When one elite group faces unusually high mobilization costs, others are less likely to assume that group will resist, thereby reducing the incentive to free ride. To test this, I examine the mechanization of the military, which has been shown to make it costlier for the military to stage coups (Caverley and Sechser, 2017; Choulis et al., 2023; Lyall and Wilson, 2009). Mechanized forces, those who rely on armored vehicles over manpower, are slow, difficult to deploy in urban environments, and ill-suited for swift, low-cost political interventions. Their deployment increases the risk of destructive confrontation and deprives officers of the element of surprise (Singh, 2014). As a result, mechanized militaries are less likely to mobilize. Recognizing this, party elites may feel compelled to act themselves, eschewing free-riding as a productive strategy.

H2: The probability of a party-led reshuffling coup attempt increases with military mobilization costs.

Data and Estimation Strategy

The unit of analysis in the models below is autocracy-year, covering all autocracies since 1946 as identified by Geddes, Wright, and Frantz (Geddes et al., 2018). The dependent variable in the analyses below is the number of attempted reshuffling coups, as identified in the Colpus dataset (Chin et al., 2021). Colpus distinguishes between reshuffling coups sponsored by military elites and those sponsored by party officials (palace coups). Importantly, I include all coup attempts, not just successful ones, as evidence of elite resistance. The goal of the analysis is not to account for the effectiveness of elite resistance, but rather for its occurrence. From this perspective, limiting the sample to successful coups would be misleading as it would effectively select on the dependent variable.

Importantly, the absence of a coup attempt could result from one of two distinct processes. On the one hand, it could be a byproduct of the coordination problem delineated above. On the other hand, it could reflect elites who are uninterested in rebelling because they are satisfied with the leader’s tenure. Most likely, the data reflect a mixture of these two processes: some non-coup cases arise from coordination failures, while others stem from elite satisfaction. It is important for the statistical models below to account for these two possibilities. Otherwise, the results would misrepresent the effect of coordination dynamics on elite resistance. I use a mixture Poisson model to address this concern. More specifically, the count of reshuffling coups attempts would be given by:

y_{i} = ZIP (p_{i}, λ_{i})

p_{i} = logistic (ϕ_{1})

λ_{i} = \exp (ϕ_{2})

The number of reshuffling coups ( $y_{i}$ ) is modeled as a zero-inflated Poisson process. The rate $λ_{i}$ represents the expected number of reshuffling coups for a given country-year. I model it as a linear function $ϕ_{1}$ primarily determined by the strength of the other elite group. The probability $p_{i}$ indicates the chance of the absence of a coup being due to satisfied elites as opposed to a coordination problem. I discuss the predictors that go into the linear functions $ϕ_{1}$ and $ϕ_{2}$ below. This estimation strategy also allows the analysis to incorporate deterrence as a competing explanation. If, for example, the military does not mobilize because a strong party deters power grabs, that dynamic will be captured into the model’s function for $p_{i}$ .

The models below incorporate a hierarchical structure. They include random effects for both linear functions, meaning that each country’s value corresponds to the average function plus a regime-level effect. The hierarchical structure is important not only because it accounts for the clustered nature of the data, but also because it allows the model to address potential overdispersion issues, as the random effects relax the assumption of equal mean and variance (Gelman and Hill, 2006). The models use Bayesian estimation to allow for more precise estimation of uncertainty (Shor et al., 2007; Stegmueller, 2013).

Following the hypotheses, I treat the rate of reshuffling coups $(λ_{i})$ as a function of military and party strength. I base my measures of elite strength on three existing indices (Gandhi and Sumner, 2020; Geddes et al., 2018; Kaire, 2024). These measures develop item-response measures that consider military and party elites simultaneously. Here, I recompile the indicators of each measure and separate them based on whether they touch on military or party strength. The former include military control over promotions and overall institutional independence (Geddes et al., 2018), absence of military purges and military involvement in government (Gandhi and Sumner, 2020; Svolik, 2012), and the extent of military fragmentation (Böhmelt et al., 2018; Kaire, 2024). The indicators for party strength include the durability of the party and its control over a unified legislature (Kaire, 2024; Templeman, 2014), a lack of cabinet changes or party purges (Banks and Wilson, 2024; Gandhi and Sumner, 2020), and whether the party retains control over its appointments (Geddes et al., 2018).

To combine the indicators above into distinct measures of military and party strength, I follow existing research and employ an item-response theory model. In particular, I reimplement the random item model developed by Kaire (2024), which has the advantage of not assuming that an indicator has the same effect across different contexts. For example, a single-party regime may indicate party strength, but not when the leader personally created the party. A random item specification allows the model to recognize this possibility and estimate regime-specific effects for each indicator. The end result is a model that does not assume structural equivalence and thus produces more precise estimates. The resulting measures are modestly correlated (r = 0.41), but an exploratory factor analysis shows that their shared eigenvalue is 0.76, suggesting that they capture distinct quantities.

In addition to elite strength, the models also account for potential confounders in $λ_{i}$ . They first control for economic performance. When times are good, elites can capture additional rents and use them to bolster their power, whereas economic crises reduce available resources and may push elites to rebel. I control for social protest for similar reasons: as popular mobilization increases, dictators often empower their elites to manage dissent. Yet this empowerment simultaneously creates opportunities for elites to oust the leader. I also include these controls in the estimation of $p_{i}$ . Overall, elites should be more satisfied when the economy is doing well, and they can avoid social unrest. In addition, I include changes in elite strength as a predictor of $p_{i}$ . When elite power declines, elites should be less satisfied with the leader and therefore more likely to rebel (but see Little, 2017). Finally, the models include a time trend to account for temporal dynamics.

Results

Table 1 presents the results of the statistical analyses, which align with the theoretical expectations: the stronger one elite is, the less likely the other is to resist the dictator. Model 1 shows that as party strength increases, the expected number of military coups decreases. Likewise, Model 2 indicates that stronger militaries are associated with fewer party coups.

Table 1.

Statistical Results.

	Model 1Military coup	Model 2Party coup	Model 3Party coup
Party strength	-2.53 (.32)^†	0.71 (0.37)	1.07 (0.47)
Military strength	1.39 (0.27)	-1.93 (0.24)^†	-0.89 (0.46)
Military Mechanization			1.61 (0.66)
Social Conflict	0.09 (.04)	-0.17 (0.15)	-0.42 (0.28)
Change in GDP	-0.09 (0.29)	-0.67 (0.42)	-0.16 (0.48)
Change in personalization	0.68 (.15)	0.68 (.15)	.79 (.05)
$σ_{m i l i t a r y}$	0.73 (0.14)
$σ_{p a r t y}$		1.01 (0.14)

Results from Bayesian multilevel zero-inflated Poisson models. N = 2,515. ^†indicates that the coefficient was estimated with country-specific effects. Standard errors in parentheses.

Figure 1 presents the predictions from Models 1 and 2. Panel A shows clear evidence of military free riding—cases where, as party power increases, the military becomes less likely to coordinate a coup. According to Model 1, when the party holds little or no power, observing at least one military coup is more likely than observing none. But if that same country were to develop strong party elites, the probability of a military coup would decline to nearly zero. This pattern is consistent with the argument that the military, recognizing that the party can now serve as a potential source of resistance, chooses to demobilize rather than bear the costs of rebellion. The results for party-led coups mirror these dynamics, as shown in Panel B: strong militaries incentivize party elites to stand aside and let the generals act.

Figure 1.

Elite Power and Incidence of Reshuffling Coups. Panel A and Panel B Show the Results from Model 1 and Model 2, Respectively.

It is also useful to analyze the extent to which the free riding equilibrium holds. It could be the case, for example, that the average effects depicted in Figure 1 could be driven by only a few cases. Indeed, we know from past research that powerful elites, on average and in isolation, discourage power grabs. Given this, Models 1 and 2 account for the potential effect heterogeneity through random slopes. The spread of these slopes, or the variance around the mean effect, are captured by $σ_{p a r t y}$ and $σ_{m i l i t a r y}$ . Both coefficients are relatively small compared to the main effect. In fact, for Model, 1 the least negative country-level coefficient for party strength is −1.8 (0.73). In Model 2, the least negative effect for military strength is −1.3 (.98). Unsurprisingly, given the rarity of elite coups, there is considerable uncertainty around this last quantity. However, the results remain clear. For most countries in the sample, the estimated effects are consistent with the free riding logic.

Model 3 tests Hypothesis 2, which posits that higher military coup costs should push party elites to resist, since it is riskier for the party to free ride when the military is unlikely to act. The results support this logic. As military mobilization costs, measured by mechanization, increase, so does the probability of a party-led reshuffling coup. The effect is substantive: a one–standard deviation increase in mechanization is associated with a 12 percent increase in the likelihood of a party-led coup. In short, free riding becomes appealing only when elites can reasonably expect others to bear the costs of resistance.

Conclusion

The argument advanced here challenges the prevailing assumption that power-sharing is uniformly stabilizing. In much of the literature, it is treated as a constraint on autocrats (Meng et al., 2023). Strong coalitions are often argued to deter dictatorial overreach because they increase the probability of coordinated backlash (Gandhi, 2008; Myerson, 2008; Svolik, 2012). However, this article suggests that there may be another side to increased power-sharing. When power is shared broadly, across many actors, coordination becomes more difficult and can create further opportunities for personalization. The consequences of power-sharing, in other words, may depend less on how much power dictators share, and more on how they distribute it.

The findings are preliminary but suggestive. They provide evidence consistent with the argument that elite coordination failures are rooted in the logic of free riding. However, the observed patterns could also be explained by elite factionalism. Elites may hesitate to resist the dictator not because they expect others to act, but because they fear that rival elites might align with the leader instead. When only one powerful elite actor exists, potential challengers can be confident that no one else will support the autocrat if a coup is attempted. But as more elites gain influence, that assurance erodes, making collective resistance riskier and less likely. Disentangling free riding from factional distrust remains an empirical challenge since both mechanisms produce the same observable implications. The results for Hypothesis 2 partially mitigate these concerns. It is not clear that factionalism can explain these results given that the disadvantages associated with mechanization are not present when the task is to shield the dictator from a palace coup. Nevertheless, future research is necessary to distinguish between these dynamics more precisely. For now, the evidence supports their shared implication: as the number of powerful elites increases, coordination becomes more difficult and opportunities for personalization expand.

Footnotes

Declaration of conflicting interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

José Kaire

Author biography

José Kaire is an Assistant Professor at the School of Politics and Global Studies of Politics, Arizona State University. His research focuses on the politics of authoritarian rule, political economy, and human rights. His book, The Road to Repression: How Neoliberalism Fuels Violence, was published by Oxford University Press and it explores some of the themes in this article.

References

Arriola

(2009) Patronage and Political Stability in Africa. Comparative Political Studies 42 (10): 1339–1362.

Banks

Wilson

(2024) Cross-National Time-Series Data (CNTS). Databanks International. https://www.cntsdata.com/.

Baturo

(2024) Personalist Regimes: A Product of Personalism and Patrimonialism. In: Baturo

Anceschi

Cavatorta

(eds) Personalism and Personalist Regimes. Oxford: Oxford University Press, pp.29–58.

Baturo

Elkink

(2021) The New Kremlinology Understanding Regime Personalization in Russia. Oxford: Oxford University Press.

Bausch

(2017) Coalition Formation and Selectorate Theory: An Experiment. Political Science Research and Methods 5 (2): 261–275.

Böhmelt

Escribà-Folch

Pilster

(2018) Pitfalls of Professionalism? Military Academies and Coup Risk. Journal of Conflict Resolution 63 (5): 1111–1139.

Boix

Svolik

(2013) The Foundations of Limited Authoritarian Government: Institutions, Commitment, and Power-Sharing in Dictatorships. Journal of Politics 53 (2): 477–494.

Bokobza

Krishnarajan

Nyrup

, et al. (2022) The Morning After: Cabinet Instability and the Purging of Ministers after Failed Coup Attempts in Autocracies. Journal of Politics 84 (3): 1437–1452.

Boutton

(2019) Coup-Proofing in the Shadow of Intervention: Alliances, Moral Hazard, and Violence in Authoritarian Regimes. International Studies Quarterly 63 (1): 43–57.

10.

Bueno

Mesquita

Smith

Siverson

, et al. (2003) The Logic of Political Survival. Cambridge, MA: The MIT Press.

11.

Casey

(2020) The Durability of Client Regimes: Foreign Sponsorship and Military Loyalty. World Politics 72 (3): 411–447.

12.

Caverley

Sechser

(2017) Military Technology and the Duration of Civil Conflict. International Studies Quarterly 61 (3): 704–720.

13.

Chin

Carter

Wright

(2021) The Varieties of Coups D’etat: Introducing the Colpus Dataset. International Studies Quarterly 65 (4): 1040–1051.

14.

Choulis

Mehrl

Escribà-Folch

, et al. (2023) How Mechanization Shapes Coups. Comparative Political Studies 56 (2): 267–296.

15.

Escribà-Folch

Timoneda

(2024) The Personalization of Power in Dictatorships. In: Lindstaedt

Den Bosch

(eds) Research Handbook on Authoritarianism. Cheltenham: Edward Elgar Publishing Limited, pp.76–95.

16.

Escribà-Folch

Böhmelt

Pilster

(2020) Authoritarian Regimes and Civil–Military Relations: Explaining Counterbalancing in Autocracies. Conflict Management and Peace Science 37 (5): 559–579.

17.

Fails

(2020) Oil Income and the Personalization of Autocratic Politics. Political Science Research and Methods 8 (4): 772–779.

18.

Gandhi

(2008) Political Institutions Under Dictatorship (Reprint edn). New York: Cambridge University Press.

19.

Gandhi

Sumner

(2020) Measuring the Consolidation of Power in Non-democracies. Journal of Politics 82 (4): 1545–1558.

20.

Geddes

Wright

Frantz

(2018) How Dictatorships Work: Power, Personalization, and Collapse. Cambridge: Cambridge University Press.

21.

Gelman

Hill

(2006) Data Analysis Using Regression and Multilevel/hierarchical Models. Cambridge: Cambridge University Press.

22.

Goemans

Gleditsch

Chiozza

(2009) Introducing Archigos: A Dataset of Political Leaders. Journal of Peace Research 46 (2): 269–283.

23.

Kaire

(2024) Can’t Coalesce, Can’t Constrain: Redefining Elite Influence in Non-democracies. Political Science Research and Methods 12 (3): 540–556.

24.

Kaire

(2025) The Road to Repression: How Neoliberalism Fuels Violence. Oxford: Oxford University Press.

25.

Kendall-Taylor

Frantz

Wright

(2017) The Global Rise of Personalized Politics: It’s Not Just Dictators Anymore. The Washington Quarterly 40 (1): 7–19.

26.

Lai

Slater

(2006) Institutions of the Offensive; Domestic Sources of Dispute Initiation in Authoritarian Regimes, 1950-1992. American Journal of Political Science 50 (1): 113–126.

27.

Little

(2017) Coordination, Learning, and Coups. Journal of Conflict Resolution 61 (1): 204–234.

28.

Lyall

Wilson

(2009) Rage against the Machines: Explaining Outcomes in Counterinsurgency Wars. International Organization 63 (1): 67–106.

29.

Magaloni

(2008) Credible Power-Sharing and the Longevity of Authoritarian Rule. Comparative Political Studies 41 (4): 715–741.

30.

Marcum

Brown

(2016) Overthrowing the ‘Loyalty Norm’: The Prevalence and Success of Coups in Small-Coalition Systems, 1950 to 1999. Journal of Conflict Resolution 60 (2): 256–282.

31.

Meng

(2020) Constraining Dictatorship: From Personalized Rule to Institutionalized Regimes. Cambridge: Cambridge University Press.

32.

Meng

Paine

Powell

(2023) Authoritarian Power Sharing: Concepts, Mechanisms, and Strategies. Annual Review of Political Science 26 (1): 153–173.

33.

Myerson

(2008) The Autocrat’s Credibility Problem and Foundations of the Constitutional State. American Political Science Review 102 (1): 125–139.

34.

Olson

(1965) The Logic of Collective Action. Cambridge, MA: Harvard University Press.

35.

Powell

Thyne

(2011) Global Instances of Coups from 1950 to 2010: A New Dataset. Journal of Peace Research 48 (2): 249–259.

36.

Reuter

(2017) The Origins of Dominant Parties: Building Authoritarian Institutions in Post-soviet Russia. Party Politics 20 (5): 663–674.

37.

Shor

Bafumi

Keele

, et al. (2007) A Bayesian Multilevel Modeling Approach to Time-Series Cross-Sectional Data. Political Analysis 15 (2): 165–181.

38.

Singh

(2014) Seizing Power: The Strategic Logic of Military Coups. Baltimore, MD: Johns Hopkins University Press.

39.

Slater

(2003) Iron Cage in an Iron Fist: Authoritarian Institutions and the Personalization of Power in Malaysia. Comparative Politics 36 (1): 81–101.

40.

Slater

(2009) Altering Authoritarianism: Institutional Complexity and Autocratic Agency in Indonesia. In: Mahoney

Thelen

(eds) Explaining Institutional Change: Ambiguity, Agency, and Power. Cambridge: Cambridge University Press, pp.132–167.

41.

Stegmueller

(2013) How Many Countries for Multilevel Modeling? A Comparison of Frequentist and Bayesian Approaches. American Journal of Political Science 57 (3): 748–761.

42.

Sudduth

(2017) Strategic Logic of Elite Purges in Dictatorships. Comparative Political Studies 50 (13): 1768–1801.

43.

Svolik

(2012) The Politics of Authoritarian Rule. Cambridge: Cambridge University Press.

44.

Templeman

(2014) Old Concept, New Cases: One-Party Dominance in the Third Wave. APSA Annals of Comparative Democratization 12 (3): 15–18.

45.

Timoneda

(2020) Institutions as Signals: How Dictators Consolidate Power in Times of Crisis. Comparative Politics 53 (1): 49–68.

46.

Woldense

(2022) What Happens When Coups Fail? The Problem of Identifying and Weakening the Enemy Within. Comparative Political Studies 55 (7): 1236–1265.