ABSTRACT
Does the perception of migration in times of crisis have implications that transcend the domain of policy support and alter systemic preferences among citizens? In an experimental study (N = 312) conducted in Macedonia, a postcommunist country on the “Balkan route”, we test how perceived threat from migrants affects attitudes towards increased concentration of political authority, conceptualised as support for presidential powers (such as veto rights, army use and policy enactment), and a presidential system. Ultimately, priming realistic and symbolic threat (in combination with presidential functions) has failed to influence participants’ attitudes, due to perceptions of low outlet credibility and high issue salience. However, subsequent analysis demonstrates that citizens who see migrants as threatening towards cultural cohesion are indeed more willing to see an empowered president. We discuss the implications of openness towards institutional design with more streamlined decision making in the context of rising illiberalism and populism in the region.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes on contributors
Ivo Bosilkov is a PhD candidate in political studies at the Network for the Advancement of Social and Political Studies (NASP) at the University of Milan, and in political communication at the Amsterdam School of Communication Research (ASCoR).
Dr. Rachid Azrout is a researcher at the Department of Communication Science at the University of Amsterdam. He recently co-authored an article named “The moderating role of identification and campaign exposure in party cueing effects”, published on 4 October 2017 in West European Politics.
ORCID
Ivo Bosilkov http://orcid.org/0000-0003-3026-2290
Rachid Azrout http://orcid.org/0000-0001-6774-3472
Notes
1. We do not use the terms “presidentialism” and “presidential powers” interchangeably. Presidentialism is a system of governance, while presidential powers are the specific powers at the president's disposal, whose extent is often, but not always determined by the institutional design of presidentialism (such as direct election, fixed term in office and control of the legislative agenda). This applies even more strongly to semi-presidental systems, such as the Macedonian one. For a detailed discussion, see Cheibub, Elkins, and Ginsburg (Citation2014). We theorise support for presidentialism as the next level – a more “pronounced” manifestation of support for presidential powers.
2. We consider it to be of little theoretical interest whether priming presidential powers would affect support for presidential powers, so we therefore move on straight to presidentialism support.
3. The association between interpersonal and political trust has been established by previous research (see for ex. Kaase Citation1999).
4. We do, however expect that trust specifically in the institution of president will increase, but only when presidential powers are primed. In that case, we expect that both the decrease of trust in other institutions, and increase of trust in the institution of president, will lead to greater support of presidential powers.
5. The alpha in the study of Reeves and Rogowski is 0.67.
6. Age (χ2 (12) = 203.08, p = n.s.), gender (χ2 (4) = 5.41, p = n.s.), ethnicity (χ2 (8) = 6.38, p = n.s.), ideology (χ2 (12) = 25.23, p = n.s.), partisanship (χ2 (24) = 15.17, p = n.s.), political knowledge (χ2 (12) = 10.82, p = n.s.).
7. Confirmatory factor analysis with unidimensional measurement for the two theoretically driven constructs had an unsatisfactory model fit (χ2 (4) = 42.84, p < .001, RMSEA = 0.20, 90% CI [.15; .25], CFI = .92), and excessively low discriminant validity (r = .95), confirming the absence of two factor solution.
8. We operationalised the IV as a dummy variable coding the non-exposure control condition (0) and the threat conditions (1).
9. We don't find any interactions effects among combinations of partisanship, political knowledge and authoritarianism, indicating that the observed divergence in partisan support isn't conditioned on political knowledge or authoritarianism levels.
10. Tests for multicollinearity indicated that despite loading on a single factor, low level of multicollinearity was present (VIF = 2.62 for MigrantEconPerspective, 2.24 for MigrantCultureImpact, 1.85 for MigrantStrainServices, 1.58 for MigrantCultureImpact and 1.36 for MigrantJobsImpact).
11. We speculate that this represents an experimental artifact, where subjects perceived the stimuli as “leading” or agenda driven.
12. However, once we include our control variables as covariates in the model, the significant interaction between authoritarianism and threat disappears.