Soonhong Cho

Peer-Reviewed Publications

  • Soonhong Cho, Doeun Kim, and Chad Hazlett. "Inference at the data's edge: Gaussian processes for estimation and inference in the face of extrapolation uncertainty" [Forthcoming, Political Analysis] [Journal Page] [arXiv] [R package (Available on CRAN)] [replication]
    • Abstract Many inferential tasks involve fitting models to observed data and predicting outcomes at new covariate values, requiring interpolation or extrapolation. Conventional methods select a single best-fitting model, discarding fits that were similarly plausible in-sample but would yield sharply different predictions out-of-sample. Gaussian Processes (GPs) offer a principled alternative. Rather than committing to one conditional expectation function, GPs deliver a posterior distribution over outcomes at any covariate value. This posterior effectively retains the range of models consistent with the data, widening uncertainty intervals where extrapolation magnifies divergence. In this way, the GP's uncertainty estimates reflect the implications of extrapolation on our predictions, helping to tame the ``dangers of extreme counterfactuals'' (King and Zeng 2006). The approach requires (i) specifying a covariance function linking outcome similarity to covariate similarity, and (ii) assuming Gaussian noise around the conditional expectation. We provide an accessible introduction to GPs with emphasis on this property, along with a simple, automated procedure for hyperparameter selection implemented in the R package gpss. We illustrate the value of GPs for capturing counterfactual uncertainty in three settings: (i) treatment effect estimation with poor overlap, (ii) interrupted time series requiring extrapolation beyond pre-intervention data, and (iii) regression discontinuity designs where estimates hinge on boundary behavior.

Working Papers

  • "Non-existent outcomes in research on inequality: A causal approach." (with Ian Lundberg) [R&R, Sociological Methods & Research] [arXiv] [R package]
    • Abstract Scholars of social stratification often study exposures that shape life outcomes. But some outcomes (such as wage) only exist for some people (such as those who are employed). We show how a common practice---dropping cases with non-existent outcomes---can obscure causal effects when a treatment affects both outcome existence and outcome values. The effects of both beneficial and harmful treatments can be underestimated. Drawing on existing approaches for principal stratification, we show how to study (1) the average effect on whether an outcome exists and (2) the average effect on the outcome among the latent subgroup whose outcome would exist in either treatment condition. We develop a framework involving parametric regression and simulation to enable principal stratification estimates that adjust for measured confounders, using standard regression-based tools. We illustrate through an empirical example about the effects of parenthood on labor market outcomes.
  • "Beyond Measurement-of-Mediation: Toward More Credible Causal Inference for Mediation-based Claims in Communication Research" (with Je Hoon Chae and Hyunjin (Jin) Song) [R&R, Human Communication Research]
    • Abstract Communication research increasingly relies on mediation to substantiate various theorized mechanisms, yet its causal evidence cannot be identified from the field’s default ``measurement-of-mediation'' designs without strong and untestable assumptions. We address this gap by introducing two research designs: (a) parallel designs that randomize both treatment and mediator, and (b) two-period crossover designs with counterbalancing and a prespecified washout interval. Using a running example from framing research, we clarify why common between-subjects designs remain insufficient, and show how these alternatives supply otherwise unavailable information for mediation-based claims. We then replicate Coleman et al. (2025) with and without these design elements to illustrate how added leverage can influence inferential precision. Finally, we discuss caveats unique to these designs and outline principled diagnostic strategies to mitigate them. By bridging the theory and practice of mediation analysis, this primer establishes a more credible pathway for causal inference for mediation-based claims in communication research.
  • "When Help Seems Optional: Institutional Projection Bias and Climate Refugee Disadvantage." (with Jieun S. Park and Margaret Peters; co-first author with Park) [Under Review]
    • Abstract Why do climate refugees receive less public support than war refugees, especially in the developed countries best positioned to assist them? We argue this reflects a cognitive mechanism we term the institutional projection bias. Those whose climate risk is buffered by robust disaster management infrastructure use their own experience as a cognitive benchmark when evaluating displacement claims, reading climate displacement as economically motivated rather than genuinely forced. Three studies test this argument. Cross-national survey evidence shows that the climate refugee disadvantage is concentrated among respondents suspicious of refugee motives, and specifically in countries with strong national disaster management capacity. Two original U.S. experiments further locate the mechanism in cognitive judgments of displacement legitimacy rather than affective prejudice, with the cause of displacement, not refugees’ origin, driving evaluations. The institutional projection bias identifies a general pattern in which effective governance inadvertently narrows public support for those who lack equivalent institutions.
  • "Let Time Tell: Identification and Gaussian Process Estimation for Interrupted Time Series." [Job Market Paper]
  • "Covariate Imbalance and Non-Overlap in Interrupted Survey Designs: A Gaussian Process Approach" (with Chad Hazlett)

Research in Progress

  • "The effects of COVID-19 Stay-at-home orders on gun purchasing." (with Jack Kappelman, Tanvi Shinkre, Ryan Baxter-King, Haotian (Barney) Chen)
    • Abstract How did the onset of the COVID-19 pandemic in the United States affect gun purchasing behavior? This paper explores the significant increase in firearms purchasing rates triggered by the pandemic and subsequent governmental restrictions to curb the virus's spread. Extant scholarship suggests that distrust in government contributed to record-breaking firearms sales, but this study uses a causal, interrupted time series design with administrative data from the National Instant Criminal Background Check System (NICS) and state-issued public health orders to provide more robust, precisely estimated effects and improved counterfactual projections. We isolate the effect of pandemic-era lockdown orders on gun purchases though our stability-controlled quasi-experimental approach, which accounts for confounding factors, data structure issues, and addresses heterogeneity in the timing of public health orders. Taken together, these results demonstrate that credible causal conclusions can be drawn even amid potential confounding. These findings deepen our understanding of the socio-political dynamics of firearm purchasing in crisis contexts and highlight the impact of government restrictions on consumer behavior, with additional findings speaking to the downstream effects this increase in gun purchasing had on firearms-involved mortality and other health outcomes.
  • "Drivers of attitudes toward Asian Americans: Disentangling stereotyping from racial identity." (with Jieun S. Park and Jessica Hyunjeong Lee)
    • Abstract Despite substantial research on attitudes toward Asian Americans, we know less about the underlying motivations. Prior studies have demonstrated that individuals make inferences about other characteristics based on someone’s race, hence feelings toward Asian Americans could be a consequence of the correlation between race and other perceived traits. To what extent are attitudes toward Asian Americans driven by dislike for racial outgroup per se, xenophobia, fear of group competition, or other presumed stereotypes? Leveraging original survey experiments that manipulate race signaled by name, immigration status, perceived threats, and stereotyped traits in short vignettes, we aim to disentangle the effects of race and other related traits inferred from it on interpersonal attitudes. Distinguishing among these competing motivations is crucial not only for understanding how perceptions of Asian Americans are shaped but also for understanding the disparate implications of interracial relationships. This study will shed light on what is the principal driver of interpersonal affect regarding race.