What happens to the producers of authoritarian propaganda when a regime falls and no lustration law is passed? To the best of my knowledge, this is the first article to study empirically the fate of propagandists after a revolution and a purge carried out without the state. I argue that journalists are ideational elites with little to bargain with. Unlike security and business elites, they command neither violence nor wealth, their credibility is questioned once the regime falls, and they lack allies powerful enough to protect them. Their fate thus depends on the distribution of power at the moment of sanction. I build new population data on Tunisian journalists from the national press catalogue. I estimate difference-in-differences and survival models against earlier cohorts, and a joint event study of the 2011 revolution and of the black book of collaborators that the presidency released in 2013. After the revolution, the exit of the ruling party’s journalists rose by 33 percentage points, whatever they had written, and few of them ever returned. Outside the party press, only the most conspicuous regime writers left. The black book came late, from files held back in the presidential palace. It named people who were mostly gone, left the others untouched and imposed its costs on the presidency. Transitions without lustration still purge, by position and not by conduct, within a window that closes quickly.
How do political assassinations affect trust in institutions in young democracies? Do citizens rally around these institutions or follow accountability cues? Political assassinations have marked new regimes and fragile democracies from Burundi to Ecuador, but the causal evidence on how publics respond comes almost entirely from established democracies and concerns attitudes toward the victim. I provide the first quasi-experimental estimate of the effect of such a killing in a new democracy on trust in state institutions. Chokri Belaid was killed in Tunisia in February 2013 while two nationally representative surveys were in the field, and opposition elites unanimously blamed the ruling Islamist party, Ennahdha. Trust in the cabinet, the parliament, and the police, all controlled by that party, rose, and trust in the army did not. Only the rise in police trust lasted through the following five weeks. The event coincides with a change of questionnaire, and a benchmark against later Tunisian survey pairs fielded without any assassination rules out a questionnaire artifact for the police and the parliament. The rally did not reflect better evaluations of the government’s performance, did not follow media salience or the ruling party’s concessions, and was weaker among partisans and university graduates. I argue that where few citizens are attached to parties, the fear that elites express across party lines outweighs the blame they assign, and it raises trust in the institutions that protect citizens more than in the government as such.
Paper presented at : MPSA, APSA, Just Research, PSRW, VIP Workshop
Much of the information necessary for political work, economic competition, and policy-making is not publicly available. Elites acquire it through informal networks, where it is brokered and gatekept rather than freely shared. This paper argues that the capacity to control such information flows, what I call informational power, constitutes a form of de facto political power that is unequally distributed across gender groups and within them. Using a novel dataset of 15,940 elite entities in Tunisia, I operationalize informational power through betweenness centrality, which captures an individual’s structural capacity to broker information between otherwise disconnected network segments. Women occupy significantly less central brokerage positions than men (beta = −0.264, p < 0.001), a finding that holds across multiple model specifications and survives controls for network size, reach, local density, and ethnic background. Comparing the gap across centrality measures locates it in brokerage: it is largest for betweenness, exceeds the degree gap in the predicted direction, and coincides with higher local clustering among women, the structural signature of ties that close existing groups in stead of bridging disconnected ones. A quantile regression and apex-tier analysis reveal that the penalty intensifies at the top of the brokerage distribution, a glass ceiling in informational power. Because the structural barriers that suppress women’s brokerage filter most women out of bridging positions entirely while a small number of exceptional brokers clear them, the between-group inequality also generates a more concentrated distribution of informational power within the female elite than within the male elite.
Paper presented at: POMEPS Workshop, APSA, POLNET, MoPED
How can researchers know who is missing from an elite interview sample? Elites de cide strategically whether to grant access, response rates depend on who the researcher is, and referral chains stay where they start, so samples can be biased in ways fieldwork never reveals. I propose a network-based protocol that maps the elite network from documentary sources before fieldwork, divides it into communities, allocates interviews in proportion to community size with at least one in each, and draws at random within communities. Unlike designs that learn the network from respondents, it takes its strata from the population’s structure before anyone is contacted. In Monte Carlo simulations, it cuts the access, degree and ideological biases of snowball sampling by more than 90% and outperforms respondent-driven sampling, multi-seed designs and network sampling with memory. On a map of Tunisia’s elite network built from seventy years of the official gazette and the business register (37,619 individuals and organi zations), referral chains reach 8 to 11 of its 52 communities and a random draw 23, while the protocol reaches up to all of them, at a cost in representativeness measurable in advance. Weighting the draw by centrality over-represents boundary-spanners and men.
Research on foreign policy attitudes treats elites as a single stratum whose interna tionalism anchors their country’s alignment with the hegemon. I argue that the social world from which a person enters the elite, not the credential that admits them, condi tions how they evaluate the actors who sustain or challenge the international order. Intramural traditional elites hold standing ascribed through proximity to the capital’s power networks; extramural rising elites reach the upper stratum through education or income from the periphery. Using Arab Barometer data from Kuwait, a compact, oil-rich, firmly US-aligned hard case, I find that traditional elites rate Putin, Erdo˘gan, Xi, Khamenei, and Assad lower than rising elites do; the groups do not differ on Biden. Proxies for sect suggest that rising elites’ warmer view of Iran and Khamenei is partly sectarian; their divergence on Assad, Putin, Erdo˘gan, and Xi is not. Nor is the divide generational: it is as large among Kuwaitis old enough to remember the 1990 Iraqi invasion as among younger ones. Education alone predicts almost nothing, and no measured attitude absorbs the divide. Rising elites also support Russia’s invasion of Ukraine less: theirs is an openness to multipolarity, not an endorsement of revisionism.
Presented at: APSA 2026
Some party systems collapse without any party gaining from the fall of another. I argue that what citizens judge governments by decides whether support rotates or drains out of the system. Reducing inequality is a demand every tradition claims, no party owns, and no government meets inside an electoral term. A verdict of distributive failure falls on whoever governs. Alternation does not discharge it, and the only exit left leads out of the party system. Trust in regimes and institutions has been studied at length, and neither can be voted out. The governing party can, and no study I know of pairs a distributive verdict with trust in it. I test the argument on Tunisia, 2011 to 2021: 6,390 plenary speeches and four Afrobarometer rounds fielded under four governments drawn from five traditions. Deputies of every family tied their confidence votes to the distributive promise of 2011. Citizens dissatisfied with the handling of income gaps trust the ruling party 0.70 points less on a four-point scale, two fifths of what satisfied citizens give. The penalty does not move with ideology. It is the same size under Islamist-led and Destourian-led coalitions, and negative in every family of the electorate. It survives fixed effects for interviewer and day, four matching estimators, and the same citizens’ evaluations of four other tasks. The design is observational, and I say what would settle the direction. To accounts of collapse built on what parties offered, I add the demand none could meet.
Presented at: APSA 2026