Topic models, as developed in computer science, are effective tools for exploring and summarizing large document collections. When applied in social science research, however, they are commonly used for measurement, a task that requires careful validation to ensure that the model outputs actually capture the desired concept of interest. In this paper, we review current practices for topic validation in the field and show that extensive model validation is increasingly rare, or at least not systematically reported in papers and appendices. To supplement current practices, we refine an existing crowd-sourcing method by Chang and coauthors for validating topic quality and go on to create new procedures for validating conceptual labels provided by the researcher. We illustrate our method with an analysis of Facebook posts by U.S. Senators and provide software and guidance for researchers wishing to validate their own topic models. While tailored, case-specific validation exercises will always be best, we aim to improve standard practices by providing a general-purpose tool to validate topics as measures.

Political borders profoundly influence outcomes central to international politics. Accordingly, a growing literature shows that historical boundaries affect important macro-outcomes such as patterns of interstate disputes and trade. To explain these findings, existing theories posit that borders have persistent effects on individual-level behavior, but the literature lacks empirical evidence of such effects. Combining spatial data on centuries of border changes in Europe with a wide range of contemporary survey evidence, we show that historical border changes have persistent effects on two of the most politically significant aspects of behavior: individuals’ political and social trust. We demonstrate that in areas where borders frequently changed, individuals are, on average, less trusting of others as well as their governments. We argue that this occurs because border changes disrupt historical state-building processes and limit the formation of interpersonal social networks, which leads to lower levels of trust.

Political scientists and policy-makers have long argued that state weakness leads to civil conflict while enhancing state power helps prevent violence. Why, then, has increased state capacity worldwide recently coincided with more civil conflicts? This study argues that enhanced state presence at the sub-national level—a symptom of growing state capacity—may induce violent resistance from the established non-state powers such as local leaders and communities in the short term. Empirically, I conduct two analyses, one at the province level and the other at the ethnic group level. To measure state presence, I use accuracy of census data in the first analysis and global ground transportation data in the second analysis. Results demonstrate that increased state presence triggers civil conflict, particularly in the first five years of such increasing state presence, and this effect is stronger in remote and ethnically heterogeneous regions. Evidence also suggests that ethnic groups settled in peripheral regions are prominent resisters to state penetration. This paper thus expands prior understanding of the role of state power in civil conflicts.

Transnational terrorism is an inherently international phenomenon as it involves attacks where the perpetrators are from a different country than the victims. Accordingly, a growing literature explains patterns in transnational attacks with a focus on international variables, for example, the presence of a border wall or alliance patterns. Despite the importance of the topic, no common empirical framework with theoretical basis has emerged to analyze the flows of transnational attacks. We propose that recent versions of the structural gravity model of transnational flows, long the workhorse model in trade economics, can be modified to provide a theoretically motivated model of the flows of transnational terrorist attacks among countries. The gravity model provides several empirical advantages for the study of international variables and transnational terrorism, for example, recent specifications allow the researcher to estimate count models that condition out all time-varying country-level confounders with fixed effects. This facilitates sidestepping the typical problem that any international variables associated with transnational flows are often correlated with omitted or imprecisely measured domestic factors, which draws their estimates into question. Moreover, we demonstrate that the structural gravity model does a much better job in predicting outcomes, particularly when multiple attacks flow across borders.