Machine Learning for Causal Inference

Abstract: Given advances in machine learning over the past decades, it is now possible to accurately solve difficult non-parametric prediction problems in a way that is routine and reproducible. In this talk, I'll discuss how these machine learning tools can be rigorously integrated into observational study analyses, and how they interact with classical ideas around randomization, semiparametric modeling, double robustness, etc. When deployed carefully, machine learning enables us to develop statistical estimators that reflect the study design more closely than basic linear regression based methods.

Race, Culture and Health: Conceptual and Methodological Innovations

Abstract: Building a culture of health and achieving health equity requires that we engage cultural processes in a more meaningful way. Cultural processes and systems are commonly referenced in health inequity scholarship but empirical research generally lags behind this conceptual emphasis. I argue that employing a transdisciplinary approach to examining intersections of culture, structure and racism is a valuable analytical tool for understanding the production of social and racial inequities in health. In this talk, I’ll discuss conceptual work advancing the concept of “cultural racism” in relation to racial inequities in health and will also provide an overview of related empirical projects: 1) a laboratory experiment examining the effects of media-based racism on physiological, psychological and behavioral stress responses, 2) a data science project exploring ways to assess chronic exposure to media-based racism and possible links to population health and 3) the use of virtual reality to promote structural competence regarding the structural and cultural roots of racism. In lieu of a deep dive on a single project or paper, the presentation seeks to support a rich conversation about the need for conceptual and methodological innovation in service of better understanding and addressing racial inequities in health.

Diversity, Immigration, and Public Opinion in the U.S.

Abstract: What factors animate public opinion towards immigrants? A substantial literature has tested the impact of individual objective traits like education and job market status on immigration attitudes. In addition, researchers have explored the role of subjective factors like immigrants’ perceived impact on society. However, prior quantitative research has generally overlooked a key aspect: natives’ impressions of who immigrants are. Immigrants in the U.S. are increasingly diverse and evidence suggests that natives prefer certain types of immigrants. Yet, survey questions gauging immigration attitudes often refer to “immigrants” as if they were a single, homogenous group, which makes it hard to interpret survey takers’ answers. To fill this gap, we explore heterogeneity in subjective perceptions of immigrants and assess their attitudinal impacts. We systematically uncover these perceptions by using a Latent Class Analysis approach on a new set of survey items we developed. We find the presence of four different immigrant “archetypes” or multidimensional constellations of immigrant traits. These archetypes are shared across regions, social classes, and partisan lines and more powerfully predict immigration attitudes than typical independent variables used in extant research. Last, we discuss the theoretical and methodological implications of our findings.

Life after death: The scale and salience of mortality in sub-Saharan Africa

Abstract: Dramatic reductions in the infant and under-five mortality rates over the last half century are among the global health community’s most notable achievements. The trends are clear and the message is positive: the world today is healthier and safer for young people than it has ever been. Sub-Saharan African countries, in particular, have experienced some of the most dramatic reductions in early life mortality. However, the all-time low infant and under-five mortality rates conceal the pervasiveness by which contemporary populations experience the phenomenon of having an infant or under-five-year-old child die—a life event that can leave parents vulnerable in myriad ways. In this talk I will introduce new population measures that capture the scale at which infant and child deaths are experienced by and dispersed across mothers in contemporary African populations. I will then demonstrate the disadvantage mothers can experience following a child’s death, and will conclude by discussing how I am extending this research with a data collection project in rural Malawi.
Co-Sponsored with the Stratification, Inequality, and Mobility Working Group

Talking Left, Living Right: Education, Ethnicity & Family Stability in the Golden State

Abstract: California has a reputation as a vanguard for the kinds of progressive values—like expressive individualism, personal fulfillment, and tolerance—associated with the second demographic transition (SDT). The SDT is associated with less marriage and greater family instability, among other things. But it turns out that, when it comes to the practice of family life, California has more intact, married families than the nation as a whole. Why is this? We argue that California has a disproportionate share of Asians and especially immigrants, and these two groups are more likely to embrace a familistic way of life and reject SDT values. We also note that more educated Californians, while they embrace progressive values in theory and in public, are more likely to embrace and live out familistic values in their own private family lives. So, immigrants, Asians, and more educated Californians disproportionately make up the ranks of Californians who are living in intact, married families.

Biodemography of Fertility and Longevity Using the Utah Population Database

Abstract: There is growing awareness that fertility affects rates of aging, adult survival prospects, and the likelihood of reaching exceptionally old ages. Much of this work, including our own, has focused on women and their ages at last birth, a proposed biodemographic marker for rates of aging. This literature has given far less attention to men, the risk of specific causes of death, the role of early initiation of fertility and how these forces may change over historical time. We use the Utah Population Database to examine how fertility alters adult mortality risks. We give special attention the role of late age at last birth, but also the role of early and late ages first birth Increasing parity is associated with worse survival for women and better for men. This talk will also present the opportunities made possible by the Utah Population Database, a unique resource of 11 million persons comprising genealogies, vital and medical records, as well as demographic and spatial data.

Does Online Partisan Media Affect Attitudes and Behavior?

Abstract: In today's fragmented online media ecosystem, does exposure to political news through partisan media have a measurable effect on citizens' political attitudes and behavior? Or are these outlets merely preaching to the choir? And if such media effects exist, are they durable and homogeneous across political groups? To answer these questions, we conducted a pre-registered, randomized field experiment embedded in a nationally representative online panel survey. We incentivized participants to temporarily alter features of their information environment during the 2018 U.S. midterm election campaign. Subjects in the treatment groups were asked to change their default browser homepage to either FoxNews.com or HuffPost.com. Using web browsing data collected for our respondents, we find that our intervention exogenously and durably altered news consumption habits. We then evaluate how our treatment affected political attitudes, voting behavior, and civic knowledge, which we measure based on survey responses collected at periodic intervals after our intervention, up to one year later. Our results generally show negligible persuasive and agenda-setting effects, consistent with the minimal media effects hypothesis. However, we uncover a meaningful decrease in overall media trust among those exposed to Fox News and an increase in support for liberal immigration policies among those in the Huffington Post treatment group.

The Effect of Land Allotment on Native American Households During the Assimilation Era

Abstract: In the early twentieth century, the U.S. government broke up millions of acres of communally owned reservation lands and allotted them to individual Native American households. Households initially received land allotments with limited property rights (‘in trust’), and were incentivized to prove themselves “competent” in order to obtain full legal title (‘fee simple’) after a set period. Indian allotment thus had elements of a conditional transfer program aimed at assimilation. The policy was ended suddenly in 1934, locking in-trust land into its status in perpetuity. We link land allotment information to the universe of Native American households in the 1940 U.S. Census. We exploit quasi-random variation in being allotted as well as in securing the allotment in fee simple. Obtaining an allotment significantly increased the likelihood of living on a farm but not of working as a farmer, indicating that allottees leased out their land. Allotments also impacted wages and occupational rank. Surprisingly, allotment most significantly impacted educational attainment. We interpret education as a way of signalling “competency” to government agents. Obtaining the land in fee simple was associated with decreased likelihood of living on a farm and owning one’s home, evidence that many allottees sold their land once they were deemed competent and obtained title. The fee-simple effects were more pronounced within tribes whose ancestral tribal norms emphasized private over communal property, indicating a cultural determinant in how the wealth transfer was utilized. Consistent with this, households in tribes with traditions of private property also engaged in more signalling of their assimilation.

Structural Racism and the People’s Health: History and Context Matters

Abstract: In this presentation on “Structural racism & the people’s health: history & context matters,” I commence with a brief reminder as to our current societal and ecological context, after which I introduce the ecosocial theory of disease distribution, which guides my work, including conceptualization and measurement of structural injustice. I then offer empirical examples of my research on structural racism and health inequities, in relation to Jim Crow and both past and present residential segregation, as measured using the Index of Concentration at the Extremes for racialized economic segregation and also historical redlining (as delineated by the 1930s federally-sponsored maps produced by the Home Owners Loan Corporation (HOLC)). Health outcomes addressed include: preterm birth; infant mortality; child mortality; cancer incidence, stage at diagnosis, and mortality; and breast cancer estrogen receptor status. The presentation concludes with reflections on embodied histories, health inequities, and the people’s health.
*Co-sponsored with the Center for the Study of Racism, Social Justice & Health

Beyond Life Expectancy—The Case for Monitoring Lifespan Variation

Abstract: Human population health is generally monitored by average mortality levels, typically in terms of life expectancies or age-standardized death rates, which belie substantial variation in length of life. Variation in ages at death, captured by a metric of lifespan variation, should be used to supplement measures of average longevity when comparing or monitoring societies and population subgroups. Although lifespan variation has historically been strongly inversely correlated with life expectancy, we are beginning to see this relationship reversed, resulting in positive correlation in some countries or subnational populations. Often these changes reflect midlife mortality crises with roots in stratified education and wealth. In this talk I will present empirical examples from around the developed world, pressing the case to monitor lifespan variation.