Researchers frequently study interactions between two distinct types of actors, naturally represented as bipartite networks. However, the existing additive and multiplicative effects (AME) framework does not accommodate bipartite data, leading to biased inference under standard dyadic regression models. This article advances the AME framework in three ways. First, I extend the model to bipartite networks. Second, I incorporate temporal dependence to capture the dynamic evolution of such networks. Third, I introduce a computationally efficient estimator based on penalized alternating least squares, replacing the original Markov Chain Monte Carlo-based approach. Using simulations and an application to global production networks, I show that the model improves coefficient estimation, more accurately recovers the data-generating process, and scales efficiently to large and complex networks. The empirical application further demonstrates its ability to uncover substantively meaningful patterns of economic relationships. I provide a R package, RAMEN, to facilitate implementation.

Foreign governments pursue influence through public diplomacy targeting citizens and direct lobbying targeting legislators, yet few studies examine these efforts jointly. This paper identifies a localized strategy of public diplomacy, grassroots diplomacy, and develops a bottom-up theory of foreign influence. I argue that grassroots diplomacy shapes public opinion, mobilizing constituent support and convincing legislators to support policies that align with foreign interests. By contrast, direct lobbying transmits information about the given state of public opinion, reassuring legislators that maintaining the status quo entails no costs and thus sustaining inaction. Using an original dataset on lobbying contacts and district-level grassroots efforts, I analyze congressional briefs in antidumping and countervailing duty investigations. I find that grassroots diplomacy promotes favorable decisions while direct lobbying prevents unfavorable ones. Grassroots diplomacy effects are stronger in districts with undecided legislators or greater information demand. These findings uncover a neglected channel of foreign influence in U.S. policymaking.

Previous research finds that elites engage the media to shape public discourse during wartime. Yet few studies examine how they contest one another within this medium. We address this by offering a theory of elite contestation and cooperation, arguing that elites seek to reinforce their group discourse while challenging rival narratives. These patterns are moderated by issue ownership. To test our claims, we compile a novel dataset of televised coverage of the Iraq and Afghan wars and construct a syntactic network to assess criticisms and endorsements among partisan and nonpartisan elites. We find that during the Iraq War, high-ranking elites were frequently criticized by out-party opponents, while co-partisans and unaffiliated experts supported them. High-casualty defeats intensified partisan criticism, moderated by issue ownership. In the Afghan War, elites criticized out-partisans but lacked consistent endorsements, and issue ownership played no significant role. Our article offers new insight into how elite wartime discourse is strategically constructed and evolves over time.

Although legislative speech is widely used to study policy preferences, few research examines how external actors, such as interest groups, influence it. We argue speech is a critical, low-cost target for lobbying influence. Using Congressional floor speeches, we estimate the effect of interest groups on speech behavior. Combining this corpus with Lobbying Disclosure Act reports (2007-2016), we find that lobbying shifts legislators' speech to be more favorable to interest groups. Moreover, lobbying that targets influential members produces significant spillover effects on rank-and-file legislators, providing new evidence of information transmission among legislators. Our analyses also show lobbying influences issue attention, specifically by reducing speech frequency---suggesting a demobilization strategy. Our text-as-data framework provides a new approach for studying how interest groups shape legislative dynamics beyond voting.