Social scientists increasingly rely on statistical models of text to resolve a wide range of questions about speech across a range of domains. However, humans communicate with more than text alone. Auditory cues convey important information, such as emotion, in many contexts of interest to social scientists. Nonetheless, researchers typically discard this information and work only with transcriptions of audio data. We develop the Structural Speaker Affect Model (SSAM), to classify auditorily distinct “modes” of speech (e.g., emotions, speakers) and the transitions between them. SSAM incorporates ridge-like regularization into a nested hidden Markov model, allowing the use of high-dimensional audio features. We implement a fast estimation procedure that enables a principled approach to uncertainty based on the Bayesian bootstrap. As a validation test, we show that SSAM markedly outperforms existing audio and text approaches in both (a) identifying individual Supreme Court justices and (b) detecting human-labeled ”skepticism” in their speech. We extend the analysis by examining the dynamics of expressed emotion in oral arguments.
We develop a novel theory as to why MPs engage in rhetorical conflict in parliamentary debates. We contend that doing so helps differentiate their party from their rivals and create a coherent and distinct party label. Given these incentives, we expect rhetorical conflict to vary with the public visibility of the debate and the status of the speaker. To test this theory, we leverage recent methodological developments in the analysis of audio and video as data. With an original corpus of New Zealand parliamentary debate videos, we use the recently developed Speaker Affect Model to measure (1) the mode of speech---either calm or conflictual---used by the speaker and (2) the presence of heckling in the chamber. Our results show that more prominent party representatives tend to deploy more conflictual speech and are the target of more heckling. Moreover, clashes are less frequent in technical stages of debate, where they may undermine cooperation on specific issues.
We built corpora of ethnographic text and audio recordings from many human societies and analyzed them with tools of quantitative social science, to explore universals and variability in music. We find that music appears in every society measured; that variation in musical behavior is best explained by three dimensions capturing the formality, excitement, and narrative importance of song events; that musical behavior varies more within societies than across societies on those dimensions; and that many common hypotheses about the behaviors regularly associated with music are supported by the ethnographic record. We demonstrate the convergent predictive validity of four different quantitative representations of music, outline the specific musical forms that universally distinguish between different song functions, and show that musical features across societies reduce to two dimensions mapping melodic and rhythmic complexity. By applying quantitative social science to rich swaths of humanistic work, the findings address longstanding debates about the nature of music.
Sectarian polarization plays a prominent role in conflicts and political movements across the Middle East, yet fundamental questions about its individual-level microfoundations remain unanswered. We consider the issues that arise in the systematic study of this phenomenon---whether and how it can be measured, the wide range of theorized causes that may drive sectarianism, its relationship to other religious and social attitudes, and the extent to which these patterns generalize---and address them in the context of an original survey of Iraqi and Iranian Shia pilgrims. Methodologically, we show that a latent-dimension measurement approach based on a novel extension of Bayesian principal component analysis produces results that are highly consistent with behavioral measures of sectarianism, self-reports, and experimental results; we then describe a unified and statistically principled approach for incorporating the numerous sources of uncertainty that arise in this type of research. Next, we integrate survey responses with a range of additional data sources to assess the observable implications of commonly theorized drivers of sectarian attitudes: economic hardship, political disillusionment, religiosity, intergroup contact, and exposure to violence. We also examine the worldview of sectarian individuals, or the extent to which sectarianism is associated with a consistent package of beliefs including sect-based political loyalties, the role of religion in government, and gender roles.
The relationship between ethnic fractionalization and lower public-goods provision is now treated as an empirical regularity. While segregated social networks are recognized as a major driver of this phenomenon, there has been little theory or evidence on the specific network dynamics that lead to differential access to public goods and services. Existing explanations have diverging implications depending on whether cross-group social ties exist at all, how easily co- and cross-group ties transmit information, and whether individuals can effectively leverage these ties to gain costly assistance. We map social networks in the highly sectarian context of contemporary Iraq, using a pool of over 300 participants from matched Baghdad neighborhoods (Sunni Adhamiya and Shia Kadhimiya). In a novel small-world network game, participants are randomly assigned to obtain information about local government offices in Sunni- or Shia-dominated target areas. We trace their search trajectories to examine how divided networks hinder citizens in their daily lives and characterize the strategies that individuals use to navigate a segregated society.
Social and medical scientists are often concerned that the external validity of experimental results may be compromised because of heterogeneous treatment effects. If a treatment has different effects on those who would choose to take it and those who would not, the average treatment effect estimated in a standard randomized controlled trial (RCT) may give a misleading picture of its impact outside of the study sample. Patient preference trials (PPTs), where participants’ preferences over treatment options are incorporated in the study design, provide a possible solution. In this paper, we provide a systematic analysis of PPTs based on the potential outcomes framework of causal inference. We propose a general design for PPTs with multi-valued treatments, where participants state their preferred treatments and are then randomized into either a standard RCT or a self-selection condition. We derive nonparametric sharp bounds on the average causal effects among each choice-based subpopulation of participants under the proposed design. We also propose a sensitivity analysis for the violation of the key ignorability assumption sufficient for identifying the target causal quantity. The proposed design and methodology are illustrated with an original study of partisan news media and its behavioral impact.