For a full list of all my publications and working papers, please see my CV.
Understanding visual messages: Visual framing and the Bag Of Visual Words
Political communication is a central element of several political dynamics. Its visual component is crucial in understanding the origin, characteristics and consequences of the messages sent between political figures, media and citizens. However, visual features have been largely overlooked in Political Science. In this project, I implement computer vision and image retrieval techniques to measure and understand messages conveyed in pictures. More specifically, the article focuses on the analysis of the content and structure of images of Black Lives Matter movement (BLM) protests. For this purpose, the article presents and details the implementation of a Bag of (Visual) Words (BoVW). This method drawn from the field of Computer Science allows researchers to build an Image-Visual Word matrix that emulates the Document-Term matrix in text analysis in order to feed models and classifiers that can provide insights about the content of visual material. Preliminary results from the application of a Structural Topic Model to a corpus of images posted by U.S. newspapers show that conservative outlets tend to include "darker" elements in their depictions of protests: they show more nocturnal events and features like smoke, fire and police patrols than liberal outlets. Overall, the article sheds light on the characteristics and consequences of visual means of communication and persuasion, and provides a useful technique for an accurate analysis and measurement of messages in pictures. (Photo credit: Robert Cohen/St. Louis Post Dispatch)
How conditioning on post-treatment variables can ruin your experiment and what to do about it
with Jacob M. Montgomery and Brendan Nyhan
Published in the American Journal of Political Science
In principle, experiments offer a straightforward method for social scientists to accurately estimate causal effects. However, scholars often unwittingly distort treatment effect estimates by conditioning on variables that could be affected by their experimental manipulation. Typical examples include controlling for post-treatment variables in statistical models or eliminating observations based on post-treatment criteria. Though these modeling choices are intended to address common problems encountered when conducting experiments, they can create severe bias in estimates of causal effects. Moreover, problems associated with conditioning on post-treatment variables remain largely unrecognized in the field, which frequently publishes studies using these practices in our discipline's most prestigious journals. We demonstrate the magnitude of the potential distortions induced by post-treatment conditioning in experiments using Monte Carlo simulations and a reanalysis of real-world data before concluding with recommendations for best practice.
Estimating controlled direct effects through marginal structural models
Conditionally accepted pending replication in Political Science Research and Methods
The estimation of direct effects allows researchers to understand causal mechanisms. However, in political science research, the treatment often affects both the confounders of the mediator and the outcome. Under these conditions traditional regression methods typically lead to two types of biases: endogenous selection and post-treatment control. This article introduces marginal structural models (MSMs) which allows unbiased estimation of controlled direct effects. I compare the inferences made from MSMs versus a naïve model using simulations and two examples regarding the effects of conditions in early stages of life on political outcomes. The analyses show that MSMs improve our understanding of causal mechanisms especially when dealing with time-varying treatments and covariates in panel and longitudinal contexts.
Learning to See: Convolutional Neural Networks for the Analysis of Social Science Data
with Francisco Cantú
This paper presents the functioning, implementation, and challenges of Convolutional Neural Networks (CNNs)---one of the most popular tools for classifying visual information. We propose that the adequate use of CNNs reduces the resources necessary for the tedious task of classifying images and extracting information from them. To illustrate the advantages and implementation of this methodology, we describe a potential application of CNNs to the collection and analysis of data on Election Day. We use this tool to code handwritten electoral results from the vote tallies of the 2015 federal election in Mexico. Our paper demonstrates the contributions of CNNs to both scholars and policy practitioners, but also presents practical challenges and limitations of the method, and provides advice on how to deal with these issues.
Through the ideology of the beholder: partisan perceptions and polarization among the mass public
with Jonathan Homola, Jon C. Rogowski, Betsy Sinclair, and Patrick D. Tucker
Scholars have documented increased polarization among contemporary mass publics but its causes and consequences are understood less well. In this paper, we study how citizens develop and apply perceptions of partisan groups. Our theoretical perspective identifies attitudinal intensity rather than partisan identities as a key accelerant of affective and social polarization. Data from two waves of a nationally representative survey demonstrate that while individuals tend to hold inaccurate perceptions of partisan out-group members, these exaggerations are driven primarily by ideological extremity rather than group membership. In a second study, we use an original survey experiment to identify the consequences of exaggerated perceptions and find that realized perceptions---but not partisanship---significantly reduced the potential for interpersonal interaction across a range of settings. Our results suggest that how partisans are perceived contribute to and reinforce social polarization and have important implications for understanding how increased party polarization affects contemporary American politics.