Interaction Models: Specification and Interpretation
Abstract
The radical interdependence between humans who live together makes virtually all human behavior conditional. The behavior of individuals is conditional upon the expectations of those around them, and those expectations are conditional upon the rules (institutions) and norms (culture) constructed to monitor, reward, and punish different behaviors. As a result, nearly all hypotheses about humans are conditional – conditional upon the resources they possess, the institutions they inhabit, or the cultural practices that tell them how to behave. Interaction Models provides a stand-alone, accessible overview of how interaction models, which are frequently used across the social and natural sciences, capture the intuition behind conditional claims and context dependence. It also addresses the simple specification and interpretation errors that are, unfortunately, commonplace. By providing a comprehensive and unified introduction to the use and critical evaluation of interaction models, this book shows how they can be used to test theoretically-derived claims of conditionality.
... 18. In so doing, I follow the advice of Berry et al. (2012) and Clark and Golder (2023) to test the conditional theory at hand as much as possible from different angles. ...
Recent research has shed light on the impact of pre-electoral coalitions on government formation in presidential democracies. However, the fact that pre-electoral coalitions are not automatically transformed into coalition cabinets has often gone under the radar. In this article, I argue that the importance of pre-electoral pacts for government formation depends on the degree of legislative polarisation. When parties are distant from one another in the ideological spectrum, presidents face more difficulties in breaking away from the pre-electoral pact and rearranging their multiparty alliances. Conversely, when polarisation is not pervasive, presidents have more leeway to build coalition cabinets different from the ones prescribed by pre-electoral coalitions. Drawing on a dataset of thirteen Latin American countries, the results support my claim and suggest that the relationship between government formation and the concession of office benefits for pre-electoral coalition members is more nuanced than previously assumed.
The literature argues that the appointment of a woman to a chief executive position is a boost for women’s empowerment in politics. Yet, while some female prime ministers (PMs) promote relatively high numbers of women among ministers, others do not. Why? What role does the PM’s gender have in defining the gender composition of the cabinet? Is ideology to play the lion’s share? This article answers through a large-N comparative analysis of 286 cabinets across Europe from 2000 to 2023, in light of an original dataset. Key findings are that no independent effect of gender and ideology can be detected. However, their interaction reveals that, due to electoral considerations, left-wing male PMs promote women more than left-wing female PMs, while there is no difference on the right. The article has implications for the debate about gendered institutions and their definition of political leaders’ strategic behavior.
Research has consistently demonstrated that Black respondents are more likely than their White counterparts to overstate their plans or previous decisions to vote. We propose that by using a single question regarding turnout intention and candidate preference (rather than the typical approach of using two separate questions), survey researchers can reduce these racial asymmetries. The reason, we suggest, is that doing so encourages Black respondents to consider the turnout question through the lens of candidates and parties, rather than as a civic responsibility they have toward other Black Americans, which therefore mitigates the social desirability bias that leads to overestimation. We evaluate this argument using a survey experiment that randomly asks a control group of survey respondents to answer two voting questions—one about turnout and one about vote choice—and asks the other half of the sample, the treatment group, a single question that combines the two. Using original data from a nationally representative survey, we find that using the single turnout/preference question reduces the racial asymmetry in turnout overestimation substantially. We further observe that the decrease is concentrated among respondents with relatively high levels of campaign indifference and especially racial civic consciousness, a condition that is much more common among Black respondents than among White respondents. We believe these findings are particularly relevant to election scholars and campaign professionals, given their focus on understanding the thought processes of survey respondents and prospective voters, as well as improving the validity of self-reports in survey research.
Countries depend on both high- and low-skilled immigration to meet economic needs. But most voters prefer high-skilled immigrants, despite the fact that multiple economic sectors structurally depend on low-skilled immigrants. In this paper, we examine voter preferences toward low-skilled immigrants as one barrier to effective immigration policy, even in political regimes where immigration is the consequence of highly coordinated or “planned” policies. Specifically, we consider whether government communication around the benefits of low-skilled immigration can increase favorability of such policies. We are particularly interested in the ways in which government communicates immigration messages and whether the scope or concentration of the proposed benefits will move individual preferences. In an online survey experiment, we present Canadians ( N=2,023) with a policy brief that manipulates immigrant skill level (high vs. low), economic outcomes of migration (positive vs. mixed), and the geographic scope of benefits (concentrated vs. sociotropic). Employing two measures of policy support, we find some evidence that positive framing can increase overall support for low-skill migrants. We also find that manipulating framing around high-skilled workers has little effect on support for low-skill workers, even when that framing presents countervailing evidence as to the benefit of high-skilled labor. In sum, our findings suggest that elite level communication around the benefits of low-skill labor may have the ability to disrupt longstanding antipathy for low-skilled labor, even in regimes with longstanding support for high-skilled labor.
ResearchGate has not been able to resolve any references for this publication.