Cha, S., Ryoo, Y., & Kim, S. E. (2023). Losing hearts and minds? Unpacking the effects of Chinese soft power initiatives in Africa. Asian Survey, 63(1), 1-30.
with Adeline Lo (University of Wisconsin-Madison, Political Science) and
Keith Levin (University of Wisconsin-Madison, Statistics)
Methods: Network analysis, spectral embedding, causal mediation analysis
Motivation: Are current causal mediation models suitable for estimating a social network effect, which requires network data as a mediator? Using proxy statistics and controlling the statistics as a confounder leads to an estimation bias and a significant amount of information loss, failing to estimate the true estimand as well.
Objectives: We introduce a novel statistical model by Alex, Frederickson, and Levin (2023), which uses network data as a mediator in causal mediation analysis. This method embeds social network data into a low-dimensional Euclidean space using singular value decomposition. By employing this embedding as a mediator in two regression models, the method effectively estimates the latent positions of actors in a social network and calculates its mediating effect.
Application: Employing this method, we show that the negative correlation between democracy and support for human rights resolutions in the UN General Assembly is potentially driven by a country's economic latent position rather than a direct effect of regime type. Improvements in democracy induce a change in a country's latent position in the inter-state economic network, and it leads to a change in voting behaviors, affected by vote-buying attempts.
with Jon Pevehouse, Junda Li, Dhavan Shah, Jisoo Kim, and Zening Duan from the Social Media And Democracy (SMAD) team at the University of Wisconsin-Madison
Methods: Topic modeling, LLM, Time series analysis, sentiment analysis
Motivation: How does social media shape public opinion on the Russo-Ukrainian War in different countries? Who are the real influencers on social media and how does the influencing structure vary across countries and through times?
Objectives: We collected extensive social media data over three platforms - Twitter (X), Reddit, and YouTube - from November 2021 to November 2022. Using structural topic modeling and time series analyses, we classify the main narratives in each platform over time and these discourses change over time. We also draw a network connection among Twitter (X) users based on their retweeting behaviors to visualize the central actors in this social media network.
Focusing on Putin's atomic diplomacy - blackmailing the Western allies with the potential use of nuclear weapons -, we explore how various narratives and events develop over time, across different platforms. Our study finds that Putin’s explicit nuclear threat rather steered public attention toward domestic politics and the partisan assessment of Wester policies.
Methods: Survey experiments, time-series analysis, text-analysis
Motivation: Despite the profound literature on democratic citizens’ high standards and moral commitment to humanitarian norms and laws, past empirical findings suggest democratic public frequently prioritizes military effectiveness or national benefits over the protection of innocent civilians abroad. Does the public show any moral concerns and drop support for military operations facing a significant number of foreign civilian casualties in inter-state conflicts? If so, why does moral pushback arise strongly in some cases but not in others?
Objectives: This study demonstrates that perceived threat levels and individuals’ primary moral foundations shape these responses, influencing whether the public opposes the war or tolerates the loss of innocent lives in foreign countries. First, individuals who prioritize individual-level moral values, such as harm/care and fairness/reciprocity, over community-level values, such as authority/respect, purity, and in-group loyalty, are more sensitive to foreign civilian casualties, leading to decreased support for war in the face of innocent deaths. However, threat perception is a critical contextual factor that can amplify the importance of community-level moral values. Using an experimental design, this paper shows that perceived threats to the in-group can temporarily heighten the salience of community-level moral foundations, thereby reducing the negative impact of foreign civilian casualties on public support for war. This study contributes to the literature on public support for war by highlighting the critical influence of moral concerns and threat perception in shaping public attitudes toward military conflicts. The findings help explain the varying reactions of democratic publics to foreign civilian deaths as time and context shift within and across conflicts.
Results: The findings support that the public perception of a violator’s intention is a critical factor affecting the public reaction to the violation of a foreign country. Among multiple factors that could shape this perception, the regime type of the violator appears to be a significant factor.
Methods: Survey experiments
Motivation: Facing numerous countries’ violations and noncompliance with international treaties, a critical question arises - How would democratic citizens react to a foreign country’s visible noncompliance with an international agreement? If the public no longer wants their government to commit to international agreements, it could lead to a diffusion of cheating and a lack of cooperation in international relations. Would the issue agenda of the treaties, the political identity of the violator, or past compliance history change the public’s reaction?
Results: The findings support that the public perception of a violator’s intention is a critical factor affecting the public reaction to the violation of a foreign country. Among multiple factors that could shape this perception, the regime type of the violator appears to be a significant factor.