Categorized
Generative Agent Simulations of 1,000 People
Joon Sung Park et al.
Stanford Working Paper, November 2024
Abstract:
The promise of human behavioral simulation -- general-purpose computational agents that replicate human behavior across domains -- could enable broad applications in policymaking and social science. We present a novel agent architecture that simulates the attitudes and behaviors of 1,052 real individuals -- applying large language models to qualitative interviews about their lives, then measuring how well these agents replicate the attitudes and behaviors of the individuals that they represent. The generative agents replicate participants' responses on the General Social Survey 85% as accurately as participants replicate their own answers two weeks later, and perform comparably in predicting personality traits and outcomes in experimental replications. Our architecture reduces accuracy biases across racial and ideological groups compared to agents given demographic descriptions. This work provides a foundation for new tools that can help investigate individual and collective behavior.
The Significance of Name-Based Racial Composition in Analyzing Neighborhood Disparities
Karl Vachuska
Socius: Sociological Research for a Dynamic World, November 2024
Abstract:
Contemporary sociological research emphasizes the need to analyze inequality beyond nominal categories. Although research has grown in this regard at the individual level, little research has pursued this approach with neighborhoods. This article explores how names can serve as a measure of the perceived typicality associated with race and how names are associated with neighborhood characteristics. Analyses on data with the names of over 300 million Americans demonstrate that name-based racial composition more fully explains socioeconomic disparities among neighborhoods than conventional survey-based racial composition metrics. Neighborhoods with the most Black-sounding names demonstrate greater socioeconomic disadvantage than neighborhoods with the most individuals self-identifying as Black. Additionally, naming patterns explain variation in socioeconomic inequality within both predominantly nominally Black neighborhoods and predominantly nominally White neighborhoods—where little nominal racial variation exists. This research suggests that infracategorical measures of race can provide additional predictive power to nominal measures of racial composition when analyzing neighborhood inequalities.
Using Information to Curb Racial Discrimination
Christopher Knittel et al.
NBER Working Paper, November 2024
Abstract:
We test whether more information about customers decreases racial bias. Our setting is the market for shared mobility services. Prior work by Ge et al. (2020) found that Uber drivers are two times more likely to cancel a ride if the passenger’s name is one used predominantly by African Americans. In a randomized control trial, we test whether two alterations to the Uber platform app reduce racial discrimination. Within the standard Uber app, drivers see only the passenger’s rating before accepting a ride. Once they accept the ride, they see the name of the passenger. In the first intervention, we increased the size of the font of the rating to draw attention to the quality of the passenger. In the second intervention, the passenger’s name appears from the beginning. Using the control group observations, we confirmed that the more likely African Americans were to use a name, the more likely a driver cancels the ride. However, increasing the font size of the passenger’s rating eliminates this racial bias. In contrast, we do not find much evidence that showing the name on the initial screen reduces or increases cancellation rates.
Household Wealth is Associated With Perceived Trustworthiness in a Diverse Set of Countries
Mélusine Boon-Falleur et al.
Social Psychological and Personality Science, forthcoming
Abstract:
Interpersonal trust impacts societal and individual outcomes, affecting economic growth, democracy, and well-being. Trust levels vary both within and across countries, raising the question of what factors influence interpersonal trust. Existing research indicates that an individual’s socioeconomic status influences their level of trust, with wealthier individuals tending to be more trusting. This article examines a further effect of wealth on interpersonal trust, namely whether people perceive wealthier individuals as more trustworthy. Using a novel method for uncovering stereotypes while avoiding social desirability bias, we investigate whether wealth cues are associated with the perceived trustworthiness of targets. Our study, conducted with diverse participants across different cultures (Brazil, Colombia, Democratic Republic of Congo, India, France, Nigeria, Philippines, and the United Kingdom), consistently demonstrates that wealthier targets are seen as more trustworthy. This culturally widespread negative stereotyping of poorer individuals may contribute to observed patterns of interpersonal trust.
Gender Stereotypes About Career and Family Are Stronger in More Economically Developed Countries and Can Explain the Gender Equality Paradox
Clotilde Napp
Personality and Social Psychology Bulletin, forthcoming
Abstract:
Using data from Project Implicit collected between 2005 and 2020, comprising 1,489,721 observations in 111 countries, we find that implicit and explicit gender stereotypes about career and family are more pronounced in more economically developed countries. Besides, these gender stereotypes are strongly correlated at the country level with gender differences in values (such as family values), self-reported personality traits (such as agreeableness or dependence), and occupational preferences (such as health-related occupations), and may account for the fact that these gender imbalances are “paradoxically” stronger in more economically developed countries (the so-called “gender equality paradox”). In line with social role theory, our findings suggest that there are in developed countries strong gender stereotypes about career and family, which may at least partly explain the persistence or even the “paradoxical worsening” of a number of gender differences in these countries, despite generally high levels of gender equality in other areas.
Some Birds Have Mixed Feathers: Bringing the Multiracial Population into the Study of Race Homophily
David Schaefer et al.
Sociological Science, November 2024
Abstract:
Research on race homophily in the United States has yet to meaningfully include the growing multiracial population. The present study confronts this challenge by drawing upon recent conceptualizations of race as a multidimensional construct. In aligning this insight with current understandings of homophily, we identify and address several open questions about the origins of race homophily -- namely regarding the possibility of peer influence on racial identity and network selection based on multiple facets of race. Data are from 3,036 youth in two large U.S. high schools with sizable proportions of mixed-race students. Using a stochastic actor-oriented model, we find that students choose friends based on similarity across multiple dimensions of racial identity and that peer influence operates to reinforce multiracial youths’ racial self-classification rather than to induce change. This points to a system where race homophily arises through multiple selection mechanisms and is reinforced by pressure toward conformity.
Psychiatric Labels and Status: Exploring Variations by Gender, Diagnosis, and Participant Attributes
Amy Kroska et al.
Socius: Sociological Research for a Dynamic World, October 2024
Abstract:
Many studies suggest that people discriminate against individuals with a mental illness. Despite these generally robust patterns, however, variability in the results from laboratory experiments examining competence-based discrimination leaves questions about the specific diagnoses that elicit the discrimination and gender differences in the discriminatory behavior. Therefore, we revisit this question with a design aimed at clarifying some of the ambiguities. We examine the effects of two diagnoses (schizophrenia and depression) and a nonpsychiatric health problem (the need for leg surgery). As with other laboratory studies, we examine if and how a teammate’s psychiatric diagnosis affects participants’ willingness to accept the teammate’s problem-solving suggestions in a two-person task group. But we go beyond the previous studies by crossing the teammate gender with participant gender and by exploring the robustness in our results by examining the moderating role of numerous participant attributes (e.g., education, social desirability, parents’ education, age, political liberalism, three gender ideology scales, trust in others). We find that participants discriminate against teammates with both types of psychiatric diagnoses but not against teammates with the nonpsychiatric health problem and that this pattern is highly robust: The processes are almost entirely unrelated to teammate gender and participant attributes, including participant gender. Together, these results suggest that both schizophrenia and depression elicit competence-based discrimination, that these processes differ very little by participants’ demographic and attitudinal attributes, and that the status beliefs underlying this discrimination may be fairly uniform.